SPEDGEI

Implementing Cisco Service

Provider Next-Generation
Edge Network Services
Version 1.0

Student Guide

Text Part Number: 97-3156-01

 Americas Headquarters Asia Pacific Headquarters Europe Headquarters
Cisco Systems, Inc. Cisco Systems (USA) Pte. Ltd. Cisco Systems International BV Amsterdam,
San Jose, CA Singapore The Netherlands

Cisco has more than 200 offices worldwide. Addresses, phone numbers, and fax numbers are listed on the Cisco Website at www.cisco.comlg%ffices.

Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this
URL: www.cisco.comigo/trademarks.Third party trademarks mentioned are the property of their respective owners. The use of the word partner does not imply a
partnership relationship between Cisco and any other company. (1110R)

DISCLAIMER WARRANTY: THIS CONTENT IS BEING PROVIDED "AS IS." CISCO MAKES AND YOU RECEIVE NO WARRANTIES
IN CONNECTION WITH THE CONTENT PROVIDED HEREUNDER, EXPRESS, IMPLIED, STATUTORY OR IN ANY OTHER
PROVISION OF THIS CONTENT OR COMMUNICATION BETWEEN CISCO AND YOU. CISCO SPECIFICALLY DISCLAIMS ALL
IMPLIED WARRANTIES, INCLUDING WARRANTIES OF MERCHANTABILITY, NON-INFRINGEMENT AND FITNESS FOR A
PARTICULAR PURPOSE, OR ARISING FROM A COURSE OF DEALING, USAGE OR TRADE PRACTICE. This learning product
rnay contain early release content, and while Cisco believes it to be accurate, it falls subject to the disclairner above.

Student Guide 2012 Cisco and/or its affiliates. All rights reserved.
 I I
I I I I
CISCO~

Students, this letter describes important

course evaluation access information!

Welcome to Cisco Systems Learning. Through the Cisco Learning Partner Program,
Cisco Systems is committed to bringing you the highest-quality training in the industry.
Cisco learning products are designed to advance your professional goals and give you
the expertise you need to build and maintain strategic networks.

Cisco relies on customer feedback to guide business decisions; therefore, your valuable
input will help shape future Cisco course curricula, products, and training offerings.
We would appreciate a few minutes of your time to complete a brief Cisco online
course evaluation of your instructor and the course materials in this student kit. On the
final day of class, your instructor will provide you with a URL directing you to a short
post-course evaluation. If there is no Internet access in the classroom, please complete
the evaluation within the next 48 hours or as soon as you can access the web.

On behalf of Cisco, thank you for choosing Cisco Learning Partners for your
Internet technology training.

Sincerely,

Cisco Systems Learning

 Table of Contents
Course Introduction 1
Overview 1
Learner Skills and Knowledge 2
Course Goal and Objectives 3
Course Flow 4
Additional References 5
Cisco Glossary of Terms 5
Your Training Curriculum 6
Your Training Curriculum 7

VPN Technologies 1-1

Overview 1-1
Module Objectives 1-1
Introducing VPNs 1-3
Overview 1-3
Objectives 1-3
VPN Concept 1-4
VPN Models 1-12
Summary 1-30
Introducing MPLS VPNs 1-31
Overview 1-31
Objectives 1-31
MPLS VPN Architecture 1-32
MPLS VPN Routing 1-42
MPLS VPN Forwarding Mechanisms 1-48
Summary 1-54
Module Summary 1-55
Module Self-Check 1-57
Module Self-Check Answer Key 1-60

MPLS Laver 3 VPNs 2-1

Overview 2-1
Module Objectives 2-1
Implementing MPLS Layer 3 VPN Backbones 2-3
Overview 2-3
Objectives 2-3
Virtual Routing and Forwarding 2-4
Example: BGP Route Propagation-Outbound 2-11
Enabling VRF 2-18
Enabling MP-BGP 2-32
Summary 2-45
Connecting Customers Using Simple Routing Protocols 2-47
Overview 2-47
Objectives 2-47
PE-CE Routing 2-48
Summary 2-66
Connecting Customers Using BGP or OSPF 2-67
Overview 2-67
Objectives 2-67
OSPF as the PE-CE Routing Protocol 2-68
BGP as the PE-CE Routing Protocol 2-83
Troubleshooting MPLS VPNs 2-94
Summary 2-101
 Deploying IPv6 and MPLS 2-103
Overview 2-103
Objectives 2-103
Support for IPv6 in MPLS 2-104
Pros 2-106
Cons 2-106
Pros 2-108
Cons 2-108
Description of Pros and Cons 2-108
Multiprotocol Extensions for BGP4 2-110
Summary 2-123
Module Summary 2-125
References 2-125
Module Self-Check 2-127
Module Self-Check Answer Key 2-129

Complex MPLS Laver 3 VPNs 3-1

Overview 3-1
Module Objectives 3-1
Implementing Complex MPLS Layer 3 VPNs 3-3
Overview 3-3
Objectives 3-3
Overlapping VPNs 3-4
Central Service VPNs 3-12
Managed CE Router Service 3-22
Summary 3-25
Implementing Internet Access and MPLS Layer 3 VPNs 3-27
Overview 3-27
Objectives 3-27
Internet Access Models with MPLS VPNs 3-28
Separate Internet Access and VPN Services 3-37
Internet Access as a Separate VPN 3-43
Summary 3-51
Introducing MPLS Interdomain Solutions 3-53
Overview 3-53
Objectives 3-53
MPLS Interdomain Solutions 3-54
CSC Models 3-63
Inter-AS 3-69
Summary 3-75
Module Summary 3-77
Module Self-Check 3-79
Module Self-Check Answer Key 3-81

Laver 2 VPNs and Ethernet Services 4-1

Overview 4-1
Module Objectives 4-1
Introducing Layer 2 VPNs 4-3
Overview 4-3
Objectives 4-3
Layer 2 VPN Overview 4-4
Summary 4-24

Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Introducing AToM 4-25
Overview 4-25
Objectives 4-25
Introduction to AToM 4-26
AToM Implementation 4-47
Summary 4-55
Implementing VPLS 4-57
Overview 4-57
Objectives 4-57
VPLS Overview 4-58
Implementing VPLS and H-VPLS 4-68
Summary 4-74
Module Summary 4-75
References 4-76
Module Self-Check 4-77
Module Self-Check Answer Key 4-78

2012 Cisco Systems, Inc. Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 iii
iv Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 SPEDGEI

Course Introduction
Overview
The Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE)
1.0 course is an instructor-led course presented by Cisco Learning Partners to their end-user
customers. This five-day course provides network engineers and technicians with the
knowledge and skills necessary to implement and support a service provider network.
The course is designed to provide service provider professionals with information on the use of
service provider VPN solutions. The goal is to train professionals to enable service provider
points of presence (POPs) to provide Layer 2 and Layer 3 VPNs. The course reinforces the
learning by providing students with hands-on labs to ensure that they thoroughly understand
how to implement VPNs within their networks.
The course also includes classroom activities with remote labs that are useful to gain practical
skills in deploying Cisco lOS, lOS XE, and lOS XR features to operate and support service
provider VPN solutions.
Learner Skills and Knowledge
This subtopic lists the skills and knowledge that learners must possess to benefit fully from the
course. The subtopic also includes recommended Cisco learning offerings that learners should
first complete to benefit fully from this course.

Students considered forthis training will have attended the following

courses or obtained equivalent-level training:
- Deploying Cisco Service Provider Network Routing (SPROUTE) v1.0
- Deploying Cisco Service Provider Advanced Network Routing
(SPADVROUTE) v1.0
- Implementing Cisco Service Provider Next-Generation Core Network Services
(SPCORE) v1.0

2 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Course Goal and Objectives
This topic describes the course goal and objectives.

To train network professionals

in the techniques to plan,
implement, and monitor
service providerVPN
solutions

~2012Cioco.ndlor~.afl'ili_.Allrighlol. . .rv"'.

Upon completing this course, you will be able to meet these objectives:
Introduce the VPN technologies that are used in the service provider environment and the
MPLS VPN peer-to-peer architecture
Describe the implementation steps that are needed to provide MPLS Layer 3 VPN service
in the service provider network
Describe how the MPLS Layer 3 VPN model can be used to implement managed services
and Internet access
Describe Layer 2 VPNs and Ethernet services

2012 Cisco Systems, Inc. Course Introduction 3

Course Flow
This topic presents the suggested flow of the course materials.

Day 1 Day2 Day3 Day 4 Day5

Course MPLS Layer 3 MPLS Layer 3 Complex MPLS Layer 2 VPNs
Introduction VPNs VPNs Layer 3 VPNs and Ethemet
(Cont.) (Cont.) Services
VPN (Cont.)
A Technologies
M

Lunch
VPN MPLS Layer 3 Complex MPLS Complex MPLS Layer 2 VPNs
Technologies VPNs Layer 3 VPNs Layer 3 VPNs and Ethemet
(Cont.) (Cont.) (Cont.) Services
(Cont.)
p
Layer 2 VPNs
M and Ethemet
Services

Cl2012Cioco.ndlor~.afl'ili_.Anrighlol. . .rv"'.

The schedule reflects the recommended structure for this course. This structure allows enough
time for the instructor to present the course information and for you to work through the lab
activities. The exact timing of the subject materials and labs depends on the pace of your
specific class.

4 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Additional References
This topic presents the Cisco icons and symbols that are used in this course, as well as
information on where to find additional technical references.

Cisco lOS Router Cisco lOS XE Router Cisco lOS XR Router

Multilayer Workgroup
Switch Switch

Network
Cloud Laptop Server

02012Cioco.ndlor~.afl'ili_.Allrighlol . . .rv"'.

Cisco Glossary of Terms

For additional information on Cisco terminology, refer to the Cisco Internetworking Terms and
Acronyms glossary of terms at
http://docwik:i.cisco.com/wiki/lnternetworking_Terms_and_Acronyms_%28ITA%29_Guide.

2012 Cisco Systems, Inc. Course Introduction 5

Your Training Curriculum
This topic presents the training curriculum for this course.

Cl2012Cioco.ndlor~.afl'ili_.Anrighlol . . .rv"'.

You are encouraged to join the Cisco Certification Community, a discussion forum open to
anyone holding a valid Cisco Career Certification (such as Cisco CCIE, CCNA, CCDA,
CCNp, CCDP(t!), CCIP(t!), CCVP, or CCSP). It provides a gathering place for Cisco certified
professionals to share questions, suggestions, and information about Cisco Career Certification
programs and other certification-related topics. For more information, visit
http://www.cisco.com/go/certifications.

6 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Your Training Curriculum
This topic presents the training curriculum for this course.

Expand Your Professional Options and Advance Your Career

WINW.cisco.com/go/certifications

~2012Cioco.ndlor~.afl'ili_.Allrighlol. . .rv"'.

2012 Cisco Systems, Inc. Course Introduction 7

8 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Module 11

VPN Technologies
Overview
This module introduces VPNs and two major VPN design options: the overlay VPN and the
peer-to-peer VPN. The module also introduces VPN terminology and topologies, and describes
Multiprotocol Label Switching (MPLS) VPN architecture and operations. This module details
various customer edge-provider edge (CE-PE) routing options and Border Gateway Protocol
(BGP) extensions (route targets and extended community attributes) that allow Internal Border
Gateway Protocol (IBGP) to transport customer routes over a provider network. The MPLS
VPN forwarding model is also covered, along with how it integrates with core routing
protocols.

Module Objectives
Upon completing this module, you will be able to describe the VPN technologies used in the
service provider environment and the MPLS VPN peer-to-peer. You will be able to meet these
objectives:
Explain the concept ofVPNs and the VPN terminology
Explain the MPLS VPN architecture, route information propagation, RDs, RTs, and virtual
routing tables
1-2 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 11

Introducing VPNs
Overview
This lesson introduces the main characteristics ofVPNs. This lesson explains the concept of
VPNs, the count advantages ofVPNs, and the VPN terminology that is also used by the
Multiprotocol Label Switching (MPLS) VPN architecture. The lesson also explains the
differences between the overlay and peer-to-peer VPN models, how they are implemented, and
the benefits and drawbacks of each implementation.
It is important to understand the background ofVPNs because you should be able to determine
when an organization might need a VPN and be able to explain how MPLS VPNs can help save
time and money. Understanding the different types ofVPNs will allow you to recognize where
they would be best used in their associated networks.

Objectives
Upon completing this lesson, you will be able to explain the concept ofVPNs and the VPN
terminology. You will be able to meet these objectives:
Describe the concept ofVPNs and the reasons why VPNs were introduced
Describe VPN implementation models, and list benefits and drawbacks ofVPNs
VPN Concept
This topic describes the concept ofVPNs and the reasons why VPNs were introduced.

Cisco NGN is a next-generation service provider infrastructure for video,

mobile, and cloud or managed services .
It provides an all-IP network for services and applications, regardless of
access type.
C2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

The Cisco IP Next-Generation Network (NGN) architecture enables service providers to start
developing fixed and mobile convergence starting with the transport in the access, aggregation,
and core networks. The NGN targets service providers with an existing centralized wireline
services edge network that are willing to maintain and evolve this network layer as part of their
services, network, and organizational evolution.
The NGN architecture provides a flexible, comprehensive, and generic framework that is
structured around the most common layers in service provider networks: customer premises,
access network, aggregation network, edge network, core network, network management, and
network admission layers. The access, aggregation, and core layers are used for transport of
mobile, video, and cloud or managed services.
The general idea of the Cisco IP NGN is to provide all-IP transport for all services and
applications, regardless of access type. IP infrastructure, service, and application layers are
separated in NGNs, thus enabling the addition of new services and applications without any
changes in the transport network.

1-4 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 IP Infrastructure Layer

VPNs relay on the IP edge and core parts of the IP infrastructure layer of
the Cisco IP NGN.

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

VPNs relay on the IP edge and core parts of the IP infrastructnre layer of the Cisco IP NGN.
Features are primarily configured on the IP edge layer. The IP core layer should be as
transparent as possible for scalability reasons..

2012 Cisco Systems, Inc. VPN Technologies 1-5

 Traditional router-based networks connect customer sites
through routers connected via dedicated point-to-point links
(leased lines).
Customer A
Leased lines

CustomerA

Traditional router-based networks were implemented with dedicated point-to-point links

connecting customer sites. The cost of this approach was comparatively high for these reasons:
The dedicated point-to-point links prevented any form of statistical infrastructure sharing
on the service provider side, resulting in high costs for the end user.
Every link required a dedicated port on a router, resulting in high equipment costs.
The complexity and cost increased by adding more remote sites.

1-6 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 -~
VPNs replace dedicated point-to-point links with emulated
point-to-point links that share common infrastructure.
Customers use VPNs primarily to reduce their operational costs.

Large Customer Site

Provider Edge (PEl Devices

custome'rr~iSi!tel==!~~J::i~~i~~4;C1rfn
Router I
Customer Premises
Equipment (CPE) or
Customer Edge (CE)

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

VPNs were introduced very early in the history of data communications with technologies such
as Frame Relay, which uses virtual circuits (VCs) to establish the end-to-end connection over a
shared service provider infrastructure. Although Frame Relay is sometimes considered
obsolete, it still shares these basic benefits with modern VPNs:
The dedicated links of traditional router-based networks have been replaced with a
common infrastructure that emulates point-to-point links for the customer, resulting in
statistical sharing of the service provider infrastructure.
Statistical sharing of the infrastructure enables the service provider to offer connectivity at
a lower price, resulting in lower operational costs for the end user.
The figure shows the statistical sharing, where the customer premises equipment (CPE) router
on the left has one physical connection to the provider edge (PE) device and two VCs have
been provisioned. VC 1 provides connectivity to the top CPE router on the right. VC 2 provides
connectivity to the bottom CPE router on the right.

2012 Cisco Systems, Inc. VPN Technologies 1-7

 3 .......
Cost savings:
- Replacing expensive long-distance leased Iines with much less expensive
dedicated connection to the service provider (OSL, fiber)
- Offloading support costs
Scalability:
- Adding a new branch office is fast and simple by adding an additional link to
the ISP (adding a site to the customerVPN).
Improved security:
- Use of encryption protocols and authentication
Better performance:
- More high-capacity data service options can be used (cheaper bandwidth).

VPNs give an organization the advantage of creating secure channels of communication while
at the same time reducing costs, improving security, increasing performance, and providing
greater access to remote users:
Cost savings: Dedicated circuits (leased lines) are quite expensive, so replacing leased lines
with a much less expensive dedicated connection to the service provider can significantly
decrease costs.
Scalability: A company with two branch offices can deploy just one dedicated line to
connect the two locations. If a third branch office needs to come online, two additional
lines will be required to directly connect that location to the other two. However, by adding
more branch offices to the network, the number ofleased lines increases dramatically (four
branch offices require six lines for full connectivity; five offices require ten lines, and so
on). VPNs avoid this problem by simply adopting one link to ISP per branch office.
Improved security: The use of encryption protocols and authentication helps secure the data
that is traveling over the VPN channel.
Better performance: VPNs also provide greater bandwidth, thus allowing better
performance.

1-8 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 I11III3-' ....... (-~
Flexibilityand reliability:
- Widespread availability offiber, OSL, and other broadband options
- Using more than one ISP
Greater access to mobile users
- Increases productivity and responsiveness for employees working from home
or on business tri ps

Flexibility and reliability: The widespread availability of fiber, DSL, and other broadband
options gives enterprises multiple ways to securely interconnect over a VPN. VPNs can
also improve reliability by using data services from several independent ISPs at the same
time (having redundant solutions).
Greater access to mobile users: Many workers can work from home or spend a significant
amount oftime on business trips. By adopting VPN solutions, they are able to connect from
anywhere to company servers to access email and data, thus increasing productivity and
responSIveness.

2012 Cisco Systems, Inc. VPN Technologies 1-9

 Large Customer Site

Provider network (P-network): The

service provider infrastructure used to
provide VPN services

Customer network (C-network): The part

of the network still under customer control

Customer site: A contiguous part of the

' - - - - I customer network (can encompass many
physical locations)

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

There are many conceptual models and terminologies that describe various VPN technologies
and implementations. The terminology is generic enough to cover nearly any VPN technology
or implementation and is thus extremely versatile.
The major parts of an overall VPN solution are always those listed here:
Provider network (P-network): The common infrastructure that the service provider uses
to offer VPN services to customers. Service provider devices to which the customer edge
(CE) routers were directly attached were called provider edge (PE) routers. In addition, the
service provider network might consist of devices used for forwarding data in the service
provider backbone called provider routers (P routers).
Customer network (C-network): The part of the overall customer network that is still
exclusively under customer control. It consists of the routers at the various customer sites.
The routers connecting the sites of individual customers to the service provider network are
called CE routers.
Customer sites: These are contiguous parts of the C-network.
A typical C-network implemented with any VPN technology would contain islands of
connectivity under customer control (customer sites) connected together via the service
provider infrastructure (P-network).

1-10 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Large Customer Site

P device: The device in the provider network with no

customer connectivity

PE device: The device in the provider network to which

the customer devices are connected

CE device: The device in the customer network that links

to the provider network (sometimes also called CPE)

PE-CE link: A link between a PE router and a CE router.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Here is a description of the devices that enable the overall VPN solution. The devices are
named based on their position in the network:
P device: Service provider devices that provide only data transport across the service
provider backbone, and have no customers that are attached to them, are called provider
devices (P devices). In traditional switched WAN implementations, these devices would be
core (or transit) switches. In an MPLS implementation, these devices would be label switch
routers (LSRs).
PE device: Service provider devices to which customer devices are attached are called PE
devices. In traditional switched WAN implementations, these devices would be Frame
Relay or X.25 edge switches. In an MPLS implementation, these devices would be edge
LSRs.
CE device: The customer router that connects the customer site to the service provider
network is called a CE router, or CE device. Traditionally, this device is called CPE.
PE-CE link: A link between a PE router and a CE router.

2012 Cisco Systems, Inc. VPN Technologies 1-11

VPN Models
This topic describes VPN implementation models, and lists benefits and drawbacks ofVPNs.

VPNs relay on the IP edge and core parts of the IP infrastructure layer of
the Cisco IP NGN.

C2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

All VPN implementation models relay on the IP edge and core parts of the IP infrastructure
layer of the Cisco IP NGN.

1-12 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 VPN services can be offered based on two major models:
Overlay model, in which the service provider provides virtual point-to-
point links between customer sites
Peer-to-peermodel, in which the service provider participates in the
customer routing

Traditional VPN implementations were all based on the overlay model, in which the service
provider sold ves between customer sites as a replacement for dedicated point-to-point links.
The overlay model had a number of drawbacks, which are identified in this lesson. To
overcome these drawbacks (particularly in IP-based customer networks), a new model called
peer-to-peer VPN was introduced. In the peer-to-peer VPN model, the service provider actively
participates in customer routing.

2012 Cisco Systems, Inc. VPN Technologies 1-13

 3
VPNs
OverlayVPN Peer-ta-Peer VPN

Layer 2 VPN Layer 3 VPN

ACLs

[ X.25 ] GRE

DMVPN
(Shared router)

Split routing
(dedicated router)

[ Frame Relay
] IPsec
[ GETVPN ]
L2TPv3

[ ATM
] SSL VPN ( MPLS VPN )
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

VPNs allow you to use the shared infrastructure of a service provider to implement your private
networks. There are basically these two implementation models:
Overlay VPNs
Layer 2, including technologies such as X.25, Frame Relay, and ATM
Layer 3, including Generic Routing Encapsulation (GRE), Dynamic Multipoint VPN
(DMVPN), IPsec, SSL VPN, and Layer 2 Tunneling Protocol (L2TP)
Peer-to-peer VPNs, implemented with routers and respective filters, separate routers per
customer via GET VPN or with MPLS VPN technology, which is covered in greater detail
in later lessons.

1-14 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 . '
. Layer 2 VPN
g g

- The service provider establishes Layer 2 VCs between customer sites.

- The customer is responsible for all higher layers.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

A Layer 2 overlay VPN implementation is the traditional switched WAN model, implemented
with technologies such as X.25, Frame Relay, or ATM. The service provider is responsible for
the transport of Layer 2 frames between customer sites, and the customer is responsible for all
higher layers.

2012 Cisco Systems, Inc. VPN Technologies 1-15

 .. ' g g
The service provider infrastructure appears as point-to-point links to the
customer.
The service provider does not see customer routes and is responsible
only for providing the point-to-point transport of customer data.
Layer 3 VPN - IP tunneling
- Routing protocols run directly
between customer routers.
IPsec SSL
- GRE is simple (and quicker).
IP
- IPsec provides authentication
and security.
Layer 2 VPN - Layer 2
forwarding
- Transparent tunneling of 802.1 Q PPP Ethernet
Layer 2 over IP
L2TPv3
IP

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

With the success of IP and associated technologies, some service providers started to
implement pure IP backbones to offer VPN services based on IP. In other cases, customers
wanted to take advantage of the low cost and universal availability of the Internet to build low-
cost private networks over it.
Whatever the business reasons behind it, Layer 3 VPN implementations over the IP backbone
always involve tunneling-encapsulation of protocol units at a certain layer of the Open
Systems Interconnection (OSI) reference model into protocol units at the same or a higher layer
of the OSI model.
Two well-known tunneling technologies are IP Security (IPsec) and GRE. GRE is fast and
simple to implement and supports multiple routed protocols, but it provides no security and is
thus unsuitable for deployment over the Internet. An alternative tunneling technology is IPsec,
which provides network layer authentication and optional encryption to make data transfer over
the Internet secure. IPsec supports only the IP routed protocol. SSL is the latest method to make
authentication and encryption of data transferred over the Internet secure. It is a remote access
solution that replaces IPsec clients and is firewall-friendly (uses SSL as the transport).
Layer 2 Tunnel Protocol Version 3 (L2TPv3) is capable oftunneling any Layer 2 payload over
L2TP.

1-16 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 The figure shows a typical Layer 2 overlay VPN implemented by a Frame Relay network.

Provider Edge (PEl Devices

CE Router - SPOKE CE Router - SPOKE

C-3 Virtual Circuits

VPN is implemented with IP-over-Frame Relay or ATM tunnels:

- The service provider establishes Layer 2 VCs between customer sites.
- The customer is responsible for all higher layers.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The customer needs to connect three sites to Site A (central site, hub site) and orders
connectivity between Site A (hub) and Site B (spoke), between Site A and Site C (spoke), and
between Site A and Site D (spoke). The service provider implements this request by providing
three permanent virtual circuits (PVCs) across the Frame Relay network, thus enabling Layer 2
connectivity between hub and spoke sites. Note that spoke-to-spoke traffic has to go through
the hub site.

2012 Cisco Systems, Inc. VPN Technologies 1-17

 Provider Edge (PEl Devices

CE Router - SPOKE CE Router - SPOKE

C-3 IP tunnels

VPN is implemented with IP-over-IP tunnels:

- Tunnels are established with GRE.
- Tunnel interfaces are point-to-point.
- Enables dynamic routing and multicast
- Runs GRE over IPsec to secure tunnel payload
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The figure presents the same scenario as the previous figure (implemented by a Frame Relay
network). The difference is that, in this case, Layer 3 connectivity is provided between hub and
spoke sites by using GRE point-to-point tunnels. The customer needs to connect three sites to
Site A (central site, hub site) and orders connectivity between Site A (hub) and Site B (spoke),
between Site A and Site C (spoke), and between Site A and Site D (spoke). The service
provider implements IP connectivity over its network. Note that spoke-to-spoke traffic has to
go through the hub site.
The GRE is a multiprotocol-capable transport protocol (IPv4, IPv6, MPLS, and so on) and
enables dynamic routing and multicast over the tunnels.
To secure the tunnel payload, you have to run GRE over IPsec.

1-18 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Provider Edge (PEl Devices

CE Router - SPOKE CE Router - SPOKE

C-3 IP tunnels e- ..,Dynamically
created IP tunnels

VPN is implemented with IP-over-IP tunnels:

- Tunnels are established with mGRE.
- Tunnel interfaces are point-to-multipoint.
- Enables dynamic routing and multicast
- Runs mGRE over IPSec to secure tunnel payload

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

In this DMVPN scenario, point-to-multipoint GRE (mGRE) tunnels are used. The customer
needs to connect three sites to Site A (central site, hub site) and orders connectivity between
Site A (hub) and Site B (spoke), between Site A and Site C (spoke), and between Site A and
Site D (spoke). The service provider implements IP connectivity over its network. Note that in
this DMVPN scenario, spoke-to-spoke traffic can flow directly by dynamically establishing
GRE tunnels between spokes. To secure the tunnel payload, you have to run mGRE over IPsec.
The GRE is a multiprotocol-capable transport protocol (IPv4, IPv6, MPLS, and so on) and
enables dynamic routing and multicast over the tunnels.

2012 Cisco Systems, Inc. VPN Technologies 1-19

 Provider Edge (PEl Devices

CE Router - SPOKE CE Router - SPOKE

C-3 IP tunnels

VPN is implemented with IP-over-IP tunnels:

- Tunnels are established with IPsec (tunnel mode).
- Enables static routing (no multicast)
- Secures payload

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

IPsec provides network layer authentication and optional encryption to make data transfer over
the Internet secure. This is achieved by creating IP-over-IP tunnels and securing the payload.
The limitation of the IPsec tunnels is that they do not offer multicast functionality, instead
providing static routing only.
The usage of IPsec, that is, securing the payload, is usually used in securing GRE and mGRE
tunnels.

1-20 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Provider Edge (PEl Devices

CE Router - SPOKE CE Router - SPOKE

C-3 L2TPv3 tunnels

L2TPv3 is used as a tunneling mechanism to deploy Layer 2 transparent

services over IP:
- L2TPv3 includes support for multiple Layer 2 encapsulations, including
802.1 Q VLAN, QinQ, and Ethernet.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Layer 2 Tunnel Protocol version 3 (L2TPv3) is capable of tunneling any Layer 2 payload over
L2TP. Specifically, L2TPv3 defines the L2TP protocol for tunneling Layer 2 payloads over an
IP core network using Layer 2 VPNs. The benefits ofthis feature include the following:
L2TPv3 simplifies deployment ofVPNs
L2TPv3 does not require MPLS
L2TPv3 supports Layer 2 tunneling over IP for any payload

2012 Cisco Systems, Inc. VPN Technologies 1-21

 Customer Site C
Provider Edge (PEl Devices

Remote-access
SSLVPN

~ ~ ...'tTERNET
~................. ~
CE Router - SPOKE '------------"----""
C-3VPN tunnels e- ..,SSL VPN
tunnel

SSL VPN enables remote-access connectivity from almost any Internet-

enabled location:
- Easy integration of the SSL VPN gateway into a shared MPLS network

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Secure Sockets Layer (SSL) is the method to achieve secure authentication and encryption of
the data transfer over the Internet. It is a remote access solution that replaces IPsec clients and
is firewall-friendly (uses SSL as the transport). It runs in three operational models:
Clientless, providing access to web servers behind the firewall
Thin client, providing port forwarding via a Java applet
Full tunnel with SSL VPN client
It is possible to integrate an SSL VPN gateway into an MPLS VPN network.

1-22 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ~--~----_.
PE-CE routing information is exchanged
between CE and PE routers.

Provider Edge (PEl Devices

PE routers exchange customer routes Customer routes are propagated through

through the core network. the PE network and sent to other CE routers.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The point-to-point overlay VPN model has a number of drawbacks, most significantly the need
for customers to establish point-to-point links or virtual circuits (VCs) between sites. The
formula to calculate how many point-to-point links or VCs are needed is ([n]*[n-l ])/2, where n
is the number of sites to be connected. For example, if a customer wants to have a full mesh
between 10 sites, it would need 10*9/2=45 point-to-point links. This would certainly be a
scalability issue.
To overcome the scalability issue and provide the customer with optimum data transport across
the service provider backbone, the peer-to-peer VPN concept was introduced. Here, the service
provider actively participates in customer routing, accepting customer routes, transporting those
customer routes across the service provider backbone, and finally propagating them to other
customer sites.

2012 Cisco Systems, Inc. VPN Technologies 1-23

 ....--~--D~1!:iiii: JI _ 1.-7)
POP router carries all customer routes.

Provider Edge (PEl Devices

Isolation between customers is

achieved with the use of ACLs (packet
filters) on PE-to-CE interfaces.

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The first peer-to-peer VPN solutions appeared with the widespread deployment of IP in service
provider networks. Architectures similar to that of the Internet were used to build these VPN
solutions. Special provisions were taken into account to transform the architecture, which was
targeted toward public backbones (Internet), into a solution in which customers would be
totally isolated and be able to exchange corporate data securely.
The more common peer-to-peer VPN implementation allowed a PE router to be shared between
two or more customers. Access control lists (ACLs-that is, packet filters) were used on the
shared PE routers to isolate the customers. In this implementation, it was common for the
service provider to allocate a portion of its address space to each customer and manage the
ACLs on the PE routers to ensure full reachability between sites of a single customer, as well as
isolation between separate customers.

1-24 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 The P router contains all
customer routes.
--_I.
Each customer has a dedicated PE
router that carries only its routes.
1

Customer Y
Site A

CE Router

Isolation between customers is

achieved through the lack of routing
information on the PE router.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Maintaining ACLs is a tedious and error-prone task. Some service providers have thus
implemented more innovative solutions based on controlled route distribution. In this approach,
the customer has a dedicated PE router. The core service P routers contain all customer routes,
and the dedicated PE routers contain only the routes of a single customer. This approach
requires a dedicated PE router per customer per point of presence (POP). Customer isolation is
achieved solely through lack of routing information on the PE router.
In the figure, the PE router for customer X, using route filtering between the P router and the
PE routers, learns only routes belonging to customer X, and the PE router for customer Y learns
only routes belonging to customer Y. Border Gateway Protocol (BGP) with BGP communities
is usually used inside the provider backbone, because it offers the most versatile route-filtering
tools.

2012 Cisco Systems, Inc. VPN Technologies 1-25

 ....--~--D~a.
Provider Edge (PEl Devices

CE Router CE Router
C-3 Payload encrypted traffic

GETVPN:
Does not use tunnels, behaves almost like transport mode IPsec
Large-scale solution accommodating multicast
Uses group security association and shared encryption key
Centralized policy and key server with periodic rekeying
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

GET VPN is a tunnel-less VPN technology that provides end-to-end security for network traffic
in a native mode and maintains the fully meshed topology. GET VPN preserves the original
source and destination IP addresses information in the header of the encrypted packet for
optimal routing (like transport mode IPsec). Hence, it is largely suited for an enterprise running
over a private MPLS and IP-based core network. It is also better suited to encrypt multicast
traffic. GET VPN uses Group Domain of Interpretation (GDOI) as the keying protocol and
IPsec for encryption. Some of the advantages of GET VPN are as follows:
Provides high scalability to any meshed topology and eliminates the need for complex
peer-to-peer security associations.
For MPLS networks, maintains network intelligence (such as full-mesh connectivity,
natural routing path, and quality of service [QoS]). Grants easy membership control with
centralized key servers.
Helps ensure low latency and jitter by enabling full-time, direct communications between
sites without requiring transport through a central hub.
Allows replication of the packets after encryption. This allows the multicast traffic to be
replicated at the core, thereby reducing the load and bandwidth requirement on the
customer premises equipment (ePE).
IP address preservation enables encrypted packets to carry the original source and
destination IP addresses in the outer IP header rather than replacing them with tunnel
endpoint addresses. This technique is known as IPsec tunnel mode with address
preservation. Some of the IP header parameters are also preserved. Many network features
like routing, basic firewall, QoS, and traffic management work based on the information
contained in the IP header. Since the IP header is persevered, all the network features will
work as before. This eliminates many issues associated with deploying point-to-point
encryption in a core network.

1-26 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Provider Edge (PE) Devices

CE routers route traffic to PE routers.

Each customer has its own isolated routing table instance on PE router.
P routers do not have customer route information.
Label switching is enabled in service provider core.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

In the MPLS VPN model, the best features of the overlay and point-to-point models are
implemented.
An MPLS-enabled core and edge network provides a very fast and efficient data switching
environment based on MPLS labels.
PE routers exchange routing information with customer CE routers and use separate isolated
routing tables for each customer. Special routing protocol contexts are used for route exchange
between PE and CE routers.
Routes are then exchanged between PE devices using the Multiprotocol BGP (MP-BGP)
routing algorithm.
For scalability reasons, service provider core routers do not have any customer routing
information. PE routers label packets with MPLS labels and P routers use these labels for fast
label-switching packets.

2012 Cisco Systems, Inc. VPN Technologies 1-27

 Overlay VPN:
- Well-known and easy to implement
- Service provider does not participate in customer routing.
- Customer network and service provider network are
well isolated.
Peer-to-peer VPN:
- Guarantees optimum routing between customersites
- Easierto provision an additional VPN
- Only sites provisioned, not links between them

Each VPN model has a number of benefits. For example, overlay VPNs have these advantages:
Overlay VPNs are well-known and easy to implement from both customer and service
provider perspectives.
The service provider does not participate in customer routing, making the demarcation
point between service provider and customer easier to manage.
On the other hand, peer-to-peer VPNs have these advantages:
They provide optimum routing between customer sites without any special design or
configuration effort.
They offer easy provisioning of additional VPNs or customer sites, because the service
provider provisions only individual sites, not the links between individual customer sites

1-28 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Overlay VPN:
- Implementing optimum routing requires a full mesh of
VCs.
- VCs have to be provisioned manually.
- Bandwidth must be provisioned on a site-to-site basis.
- Overlay VPNs always incur encapsulation overhead (GRE or IPsec).
Peer-to-peer VPN:
- The service provider participates in customer routing. Filters should be applied
to customer links.
- The service provider becomes responsible for customer convergence.
- P E routers carryall routes from all custom ers.
- A secu re envi ronment must be provided for customers.
- Complex configuration
- The service provider needs detailed IP routing knowledge.

Each VPN model also has a number of drawbacks. Overlay VPNs have these disadvantages:
Overlay VPNs require a full mesh of VCs between customer sites to provide optimum site-
to-site routing.
All VCs between customer sites must be provisioned manually, and the bandwidth must be
provisioned on a site-to-site basis (which is not always easy to achieve).
The IP-based overlay VPN implementations (with IPsec or GRE) incur high encapsulation
overhead-ranging from 20 to 80 bytes per transported datagram.
The major drawbacks of peer-to-peer VPNs arise from service provider involvement in
customer routing, such as these disadvantages:
The service provider becomes responsible for correct customer routing and for fast
convergence of the C-network following a link failure.
The service provider PE routers need to carryall customer routes that were hidden from the
service provider in the overlay VPN model.
The service provider needs detailed IP routing knowledge, which is not readily available in
traditional service provider teams.
PE routers have more complex configuration.
A secure environment must be provided for customers.

2012 Cisco Systems, Inc. VPN Technologies 1-29

Summary
This topic summarizes the primary points that were discussed in this lesson.

Two options:
- Traditional router-based networks connect via dedicated point-to-point links.
- VPNs use emulated point-to-point links sharing a common infrastructure.
The two majorVPN models are overlayVPN and peer-to-peerVPN:
- Overlay VPNs use well-known technologies and are easy to im plement.
- Overlay VPN VCs have to be provisioned manually.
- Peer-to-peer VPNs guarantee optimum routing between customersites.
- Peer-to-peer VPNs require that the service provider participate in customer
routi ng.

1-30 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 21

Introducing MPLS VPNs

Overview
This lesson explains the Multiprotocol Label Switching (MPLS) VPN architecture. It is
important to understand how the MPLS VPN architecture is structured, what the components of
that architecture are, and how the components are used. This knowledge will help later, when
you begin to look at design issues and configuration parameters. The lesson offers address and
routing perspectives from the customer and service provider side, and it discusses how routing
tables appear on provider edge (PE) routers. This lesson also explains how forwarding across
an MPLS VPN backbone occurs, identifies how labels get propagated, and explains the effects
of summarization in the core.

Objectives
Upon completion of this lesson, you will be able to understand MPLS VPNs. You will be able
to meet these objectives:
Explain the MPLS VPN architecture, RDs, RTs, and virtual routing tables
Describe end-to-end routing update flow
Describe VPN label propagation between PE routers and the MPLS VPN end-to-end
forwarding mechanism
MPLS VPN Architecture
This topic explains the MPLS VPN architecture, route distinguishers (RDs), route targets
(RTs), and virtual routing tables.

An MPLS VPN combines the best features of an overlay VPN

and a peer-to-peer VPN:
PE routers participate in customer routing, guaranteeing optimum
routing between sites and easy provisioning.
PE routers carry a separate set of routes for each customer (similar to
the dedicated PE router approach).
Customers can use overlapping addresses.

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

In the MPLS VPN architecture, the edge routers carry customer routing information, providing
optimal routing for traffic that belongs to the customer for intersite traffic. The MPLS-based
VPN model also accommodates customers who use overlapping address spaces, unlike the
traditional peer-to-peer model, in which optimal routing of customer traffic required the
provider to assign IP addresses to each of its customers (or the customer to implement Network
Address Translation [NAT]) to avoid overlapping address spaces. MPLS VPN is an
implementation of the peer-to-peer model; the MPLS VPN backbone and customer sites
exchange Layer 3 customer routing information, and data is forwarded between customer sites
using the MPLS-enabled service provider IP backbone.

1-32 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customer A Customer A
Site 1 Site 2
CE1-A
CE Router

P1
PE1 PE2
MPLS VPN Service
Provider Edge Provider Network
Provider Edge

"",,~'
Router Router
P2 \
V\ Customer B
\
\
Site 2

CE1-B CE2-B--"fI!!!I
CE Router CE Router

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

The MPLS VPN domain, like the traditional VPN, consists of the customer network and the
provider network. The MPLS VPN model is very similar to the dedicated provider edge (PE)
router model in a peer-to-peer VPN implementation. However, instead of deploying a dedicated
PE router per customer, customer traffic is isolated on the same PE router that provides
connectivity into the service provider network for multiple customers.
The main components ofMPLS VPN architecture are as follows:
Customer network: Usually a customer-controlled domain consisting of devices or routers
spanning multiple sites that belong to the customer.
Customer edge (CE) routers: Routers in the customer network that interface with the
service provider network.
Provider network: The provider-controlled domain consisting of PE and provider core
routers that connect sites belonging to the customer on a shared infrastructure. The provider
network controls the traffic routing between sites belonging to a customer, along with
customer traffic isolation.
PE routers: Routers in the provider network that interface or connect to the customer edge
routers in the customer network.
P routers: Routers in the core of the provider network that interface with either other
provider core routers or provider edge routers.

Note In an MPLS VPN implementation, the PE router is the edge label switch router (edge LSR).

2012 Cisco Systems, Inc. VPN Technologies 1-33

 A PE router in an MPLS VPN uses virtual routing tables to implement
the functionality of customer-dedicated PE routers.

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

An MPLS VPN implementation is very similar to a dedicated router peer-to-peer model

implementation. From the perspective of a CE router, only IPv4 updates, as well as data, are
forwarded to the PE router. The CE router does not need any specific configuration to enable it
to be a part of an MPLS VPN domain. The only requirement on the CE router is a routing
protocol (or a static default route) that enables the router to exchange IPv4 routing information
with the connected PE router.
In the MPLS VPN implementation, the PE router performs multiple functions. The PE router
must first be capable of isolating customer traffic if more than one customer is connected to the
PE router. Each customer, therefore, is assigned an independent routing table (virtual routing
table or virtual routing and forwarding [VRF] table) similar to a dedicated PE router in the
initial peer-to-peer discussion. Routing across the service provider backbone is performed using
a routing process in the global routing table. P routers provide label switching between PE
routers and are unaware ofVPN routes. CE routers in the customer network are not aware of
the P routers, and thus the internal topology of the service provider network is transparent to the
customer. The figure shows the functionality of the PE router.
The P routers are responsible only for label switching of packets. They do not carry VPN routes
and do not participate in MPLS VPN routing. The PE routers exchange IPv4 routes with
connected CE routers using individual routing protocol contexts. To enable scaling the network
to a large number of customer VPNs, Multiprotocol Border Gateway Protocol (MP-BGP) is
configured between PE routers to carry customer routes.

1-34 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customer A Site 1
Global Routing

-.... .......... ......... .. ....,

Table
CE1-A Routes

CE1-A
~---_
- ....
.... Virtual Routing Table
(Customer A)

Virtual Routing Table

CE1-B ...
_... ... ...
~
; (Customer B)

--
...... CE1-B Routes
...
... ...
PE Router

...
Customer B Site 1

The VRF contains an IP routing table that is analogous to the global IP routing table, a Cisco
Express Forwarding table, a list of interfaces that are part of the VRF, and a set of rules
defining routing protocol exchange with attached CE routers (routing protocol contexts). In
addition, the VRF also contains VPN identifiers and VPN membership information (route
distinguisher [RD] and route target [RT] are covered in the next section). The interface that is
part of the VRF must support Cisco Express Forwarding switching. The number of interfaces
that can be bound to a VRF is limited only by the number of interfaces on the router, and a
single interface (logical or physical) can be associated with only one VRF.
The figure shows the function of a VRF on a PE router to implement customer routing
isolation. Cisco lOS Software supports various routing protocols and individual routing
processes (Open Shortest Path First [OSPF], Enhanced Interior Gateway Routing Protocol
[EIGRP], and so on) per router. However, for some routing protocols, such as Routing
Information Protocol (RIP) and Border Gateway Protocol (BGP), Cisco lOS Software supports
only a single instance of the routing protocol. Therefore, to implement per-VRF routing using
protocols that are completely isolated from other VRFs, which might use the same provider
edge-customer edge (PE-CE) routing protocols, the concept of routing context was developed.

2012 Cisco Systems, Inc. VPN Technologies 1-35

 A dedicated routing protocol is used
to carry customer routes between PE routers.

p
Router

P-Network
CustomerC CustomerC

Run a single routing protocol that will carry all customer routes between
PE routers. Use MPLS labels to exchange packets between PE routers.
P routers do not carry customer routes; the solution is scalable.
The number of customer routes can be very large. BGP4 is the only
routing protocol that can scale to a very large number of routes.
BGP is used to exchange customer routes directly between PE routers.
Extend the customer addresses to make them unique.
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Although virtual routing tables provide isolation between customers, the data from these
routing tables still needs to be exchanged between PE routers to enable data transfer between
sites attached to different PE routers. Therefore, a routing protocol is needed that will transport
all customer routes across the P-network while maintaining the independence of individual
customer address spaces.
The best solution to the customer route propagation issue is to run a single routing protocol
between PE routers that will exchange all customer routes without the involvement of the P
routers. This solution is scalable. Some of the benefits of this approach are as follows:
The number of routing protocols running between PE routers does not increase with an
increasing number of customers.
The P routers do not carry customer routes.
The next design decision to be made is the choice of the routing protocol running between PE
routers. Given that the total number of customer routes is expected to be very large, the only
well-known protocol with the required scalability is Border Gateway Protocol version 4
(BGP4). In fact, BGP4 is used in the MPLS VPN architecture to transport customer routes
directly between PE routers.
MPLS VPN architecture differs in an important way from traditional peer-to-peer VPN
solutions: MPLS VPNs support overlapping customer address spaces.
With the deployment of a single routing protocol (that is, BGP4) exchanging all customer
routes between PE routers, an important issue arises: how can BGP4 propagate several identical
prefixes, belonging to different customers, between PE routers?
The only solution to this dilemma is the expansion of customer IP prefixes with a unique prefix
that makes them unique even if they had previously overlapped. A 64-bit prefix called the route
distinguisher (RD) is used in MPLS VPNs to convert non-unique 32-bit customer addresses
into 96-bit unique addresses that can be transported between PE routers.

1-36 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 In the MPLS VPN backbone, the PE router needs to implement
processes that enable overlapping address spaces in connected
customer networks.
The 64-bit route distinguisher is prepended to an IPv4 address to make
it globally unique.
The resulting address is a VPNv4 address.
VPNv4 addresses are exchanged between PE routers via BGP4.

Route Distinguisher IPv4 Address VPNv4

(8 Bytes) (4 Bytes) Address

",,
,,
,
RD Formats {II; AS Number
VPN Identifier I
IPAddress VPN Identifier I

In the MPLS VPN routing model, the PE router provides isolation between customers using
VRFs. However, this information must be carried between PE routers to enable data transfer
between customer sites via the MPLS VPN backbone. The PE router must be capable of
implementing processes that enable overlapping address spaces in connected customer
networks. The PE router must also learn these routes from attached customer networks and
propagate this information using the shared provider backbone. This is done by the association
of an RD per virtual routing table on a PE router.
An RD is a 64-bit unique identifier that is prepended to the 32-bit customer prefix or route
learned from a CE router, which makes it a unique 96-bit address that can be transported
between the PE routers in the MPLS domain. Thus, a unique RD is configured per VRF on the
PE router. The resulting address, which is 96 bits total (32-bit customer prefix plus 64-bit
unique identifier or RD), is called a VPNv4 address.
VPNv4 addresses are exchanged only between PE routers; they are never used between CE
routers. Between PE routers, BGP must therefore support the exchange of traditional IPv4
prefixes and the exchange ofVPNv4 prefixes. A BGP session between PE routers is therefore
called a Multiprotocol Border Gateway Protocol (MP-BGP) session.
The format of an RD is shown in the figure. An RD can be of two formats. If the provider does
not have a BGP autonomous system (AS) number, the IP address format can be used, and if the
provider does have an AS number, the AS number format can be used.

2012 Cisco Systems, Inc. VPN Technologies 1-37

 96-bit VPNv4 Prefix
1- - - - RD: 1:100:172.16.10.0/24 - - - ,
CustomerA 1 r-"RD:1:101:172.16.10.0/24-1 I
Site 1 r;=====::::::;l VPNv4Prefix r.=====:::::;l

PE Router PE Router

CustomerB '--------------" CustomerB

Site 1 Site 2

In the figure, the same IP prefix, 172.16.10.0/24, received from two different customers, is
made unique by prepending different RD values, 1: 100 and 1: 101, before propagating the
addresses as VPNv4 addresses on the PE router.
The protocol used for exchanging these VPNv4 routes between PE routers is MP-BGP; BGP
that is capable of carrying VPNv4 (96-bit) prefixes in addition to other address families is
called MP-BGP. The IGP requirement to implement Internal Border Gateway Protocol (IBGP)
still holds in the case of an MPLS VPN implementation. Therefore, the PE router must run an
IGP that provides Network Layer Reachability Information (NLRI) for IBGP ifboth PE routers
are in the same AS.
MP-BGP is also responsible for the assignment of a VPN label. Packet forwarding in an MPLS
VPN mandates that the router specified as the next hop in the incoming BGP update is the same
router that assigns the VPN label.
An MP-BGP session between PE routers in a single BGP AS is called a Multiprotocol Internal
Border Gateway Protocol (MP-IBGP) session and follows rules as in the implementation of
IBGP with regards to BGP attributes. If the VPN extends beyond a single AS, VPNv4 routes
will be exchanged between autonomous systems at the AS boundaries using a Multiprotocol
External Border Gateway Protocol (MP-EBGP) session.

1-38 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customer A Customer A
Site 1 CE2-A Site 2
CE Router ",:!Iiii~"",

PE1
Provider Edge
Router

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

Customer route propagation across an MPLS VPN network is done using this process:
Step 1 The CE router sends an IPv4 routing update to the PE router.
Step 2 The PE router prepends a 64-bit RD to the IPv4 routing update, resulting in a
globally unique 96-bit VPNv4 prefix.
Step 3 The VPNv4 prefix is propagated via an MP-IBGP session to other PE routers.
Step 4 The receiving PE routers strip the RD from the VPNv4 prefix, resulting in an IPv4
prefix. RD is used to match the proper VRF routing table.
Step 5 The IPv4 prefix is forwarded to other CE routers within an IPv4 routing update.

2012 Cisco Systems, Inc. VPN Technologies 1-39

 The RD cannot identify participation in more than one VPN.
RTs were introduced in the MPLS VPN architecture to support complex
VPN topologies.
RTs are additional attributes attached to VPNv4 BGP routes to indicate
VPN membership.
Extended BGP communities are used to encode these attributes.
Export RTs:
- IdentifyingVPN membership
- Appended to the customer route when it is converted into a VPNv4 route
Import RTs:
- Associated with each vi rtual routi ng table
- Select routes to be inserted intothe virtual routing table

Consider a scenario where some sites have to participate in more than one VPN. In such a
scenario, the RDs cannot identify participation in more than one VPN.
Route targets (RTs) were introduced into the MPLS VPN architecture to support identifying a
site that participates in more than one VPN. RTs are additional identifiers used in the MPLS
VPN that identify the VPN membership of the routes learned from that particular site. RTs are
implemented by the use of extended BGP communities in which the higher-order 16 bits of the
BGP extended community (64 total bits) are encoded with a value corresponding to the VPN
membership of the specific site. When a VPN route learned from a CE router is injected into
VPNv4 BGP, a list ofVPN route target extended community attributes is associated with it.
MPLS VPN RTs are attached to a customer route at the moment that it is converted from an
IPv4 route to a VPNv4 route by the PE router. The RTs attached to the route are called export
RTs and are configured separately for each virtual routing table in a PE router. Export RTs
identify a set ofVPNs in which sites associated with the virtual routing table belong.
When the VPNv4 routes are propagated to other PE routers, those routers need to select the
routes to import into their virtual routing tables. This selection is based on import RTs. Each
virtual routing table in a PE router can have a number of configured import RTs that identify
the set ofVPNs from which the virtual routing table is accepting routes.
When implementing complex VPN topologies, such as extranet VPN, Internet access VPNs,
network management VPN, and so on, using MPLS VPN technology, the RT plays a pivotal
role. A single prefix can be associated to more than one export route target when propagated
across the MPLS VPN network. The RT can, as a result, be associated to sites that might be a
member of more than one VPN.

1-40 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 1:100:172.16.10.0/24
RT1:100 NH 10.10.10.101 (PE1) VPN Label:V1
1:101 :192.168.1 0.0/24
RT1:101 NH 10.10.10.101 (PE1) VPN Label:V2

Customer A tt: : Customer A

Site 1 Site 2
MP-BGP [ MP-BGP
,/~--
/
"" VRF Customer A 1/
\
VRF Customer A
; "" """" CE1-A''-J'- RD = 1:100 =
RD 1:100
( =
Export RT 1:100 Export RT = 1:100
Import RT = 1:100 =
Import RT 1:100

VRF Customer B VRF Customer B

=
RD 1:101 =
RD 1:101
Export RT = 1:101 Export RT = 1:101
=
Import RT 1:101 =
Import RT 1:101 1''-. CE2-B

PE Router (PE1) PE Router (PE2)

'----- ---' MPLS VPN '----- ---'
Customer B Service Customer B
Site 1 Provider Site 2
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The following processes occur during route propagation in an MPLS VPN, as shown in the
figure:
1. The prefix 172.16.10.0/24 is received from CEl-A, which is part ofVRF Customer A on
PEL The prefix 192.168.10.0/24 is received from CE1-B, which is part ofVRF Customer
B on PEL
2. For CE I-A, the PE 1 associated an RD value of 1: 100 and an export RT value of 1: 100, as
configured in the VRF definition on the PE router. For CEl-B, the PEl associated an RD
value of 1: 101 and an export RT value of 1: 101.
3. Routes learned from connected CE routers CE1-A are redistributed into the MP-BGP
process on PE, where the prefix 172.16.10.0/24 is prepended with the RD value of 1:100
and appended with the route target extended community value (export RT) of 1: 100 prior to
sending the VPNv4 prefix as part of the MP-IBGP update between PE routers.
Routes learned from connected CE routers CE1-B are redistributed into the MP-BGP
process on PEl, where the prefix 192.168.10.0/24 is prepended with the RD value of 1: 10 1
and appended with the route target extended community value (export RT) of 1: 101 prior to
sending the VPNv4 prefix as part of the MP-IBGP update between PE routers.
The VPN label (4 bytes) is assigned for each prefix that is learned from the IGP process of
the connected CE router within a VRF by the MP-BGP process of the PE router. MP-BGP
running in the service provider MPLS domain thus carries the VPNv4 prefix (IPv4 prefix
with prepended RD) in addition to the BGP route target extended community.
The next hops on PE routers must not be advertised in the BGP process but must be learned
from the IGP for MPLS VPN implementation. The VPN label is depicted by the entries VI
and V2 in the figure.
4. The MP-BGP update is received by the PE router PE2, and the route is stored in the
appropriate VRF table for Customer A, based on the VPN label.
5. The received MP-BGP routes are redistributed into the VRF PE-CE routing processes, and
the route is propagated to CE2-A.

2012 Cisco Systems, Inc. VPN Technologies 1-41

MPLS VPN Routing
This topic describes the end-to-end routing update flow for MPLS VPNs.

CE routers must run standard IP routing software.

PE routers must support MPLS VPN services and IP routing.
P routers must not participate in customerVPN routing.

The designers of MPLS VPN technology were faced with these routing requirements:
CE routers should not be MPLS VPN-aware; CE routers should run standard IP routing
software.
PE routers must support MPLS VPN services and traditional Internet services.
To make the MPLS VPN solution scalable, P routers must not participate in customer VPN
routing. P routers use only label switching.

1-42 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Exchange VPN routes with CE routers via per-VPN routing protocols
Exchange core routes with P routers and PE routers via core IGP

P routers do not participate in MPLS VPN

The CE routers run standard IP routing software and routing and do not carry VPN routes.
exchange routing updates with the PE router.
P routers run backbone IGP with the PE routers
The PE router appears as another router in the C-network. and exchange information about global
subnetworks (core links and loopbacks).
The P routers are hidden from the customer.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The MPLS VPN backbone should look like a standard corporate backbone to the CE routers.
The CE routers run standard IP routing software and exchange routing updates with the PE
routers, which appear to them as normal routers in the customer network (C-network). An
MPLS VPN implementation is very similar to a dedicated router peer-to-peer model
implementation. From the perspective of a CE router, only IPv4 updates and data are forwarded
to the PE router. The CE router does not need any specific configuration to enable it to be a part
of an MPLS VPN domain. The only requirement on the CE router is a routing protocol (or a
static default route) that enables the router to exchange IPv4 routing information with the
connected PE router, which appears as normal router in the C-network.
From the customer perspective, the MPLS VPN backbone looks like an intracompany BGP
backbone with PE routers performing route redistribution between individual sites and the core
backbone. The standard design rules that are used for enterprise BGP backbones can be applied
to the design of the C-network. The P routers are hidden from customer view; the interual
topology of the BGP backbone is transparent to the customer.
From the P router perspective, the MPLS VPN backbone looks even simpler-the P routers do
not participate in MPLS VPN routing and do not carry VPN routes. The P routers run only a
backbone IGP with other P routers and with PE routers, and exchange information about core
subnetworks. BGP deployment on P routers is not needed for proper MPLS VPN operation;
BGP deployment might be needed, however, to support traditional Internet connectivity that
has not yet been migrated to MPLS.
The PE routers are the only routers in the MPLS VPN architecture that see all routing aspects
of the MPLS VPN. PE routers can perform these exchanges:
PE routers exchange IPv4 VPN routes with CE routers via various routing protocols
running in the virtual routing tables.
PE routers exchange VPNv4 routes via MP-IBGP sessions with other PE routers.
PE routers exchange core routes with P routers and other PE routers via core IGP.

2012 Cisco Systems, Inc. VPN Technologies 1-43

 ~"--5
PE routers can run standard IPv4 BGP in the global routing
table:
PE routers exchange Internet routes with other PE routers.
CE routers do not participate in Internet routing.
P routers do not need to participate in Internet routing.

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The routing requirements for PE routers also extend to supporting Internet connectivity-PE
routers need to exchange Internet routes with other PE routers. The CE routers cannot
participate in Internet routing if the Internet routing is performed in global address space. The P
routers could participate in Internet routing; however, Internet routing should be disabled on the
P routers to make the network core more stable.

1-44 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 PE routers contain a number of routing tables:
The global routing table contains core routes (filled with core IGP) and
Internet routes (filled with IPv4 BGP) .
The VRF tables contain routes for sites of identical routing requirements
from local (IPv4 VPN) and remote (VPNv4 via MP-BGP) CE routers.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The PE routers fulfill various routing requirements imposed on them by using a number of IP
routing tables. Here are some examples:
The global IP routing table (the IP routing table that is always present in a router even if it
is not supporting an MPLS VPN) contains all core routes (inserted by the core IGP) and
Internet routes (inserted from the global IPv4 BGP table).
The VRF tables contain sets of routes for sites with identical routing requirements. The
VRFs are filled with intra-VPN IGP information exchanged with the CE routers and with
VPNv4 routes received through MP-BGP sessions from the other PE routers.

2012 Cisco Systems, Inc. VPN Technologies 1-45

 --~--_.
PE routers receive IPv4 routing
updates from CE routers and install
them in the appropriate VRF table.

PE routers export VPN routes from VRF

CE Router tables into MP-BGP and propagate them as CE Router
VPNv4 routes to other PE routers. The
export RT attribute is matched.

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The figure shows how PE routers receive IPv4 routing updates from the CE routers and install
the updates in the appropriate VRF table. Cisco lOS, IOS XE, and lOS XR Software currently
supports RIP version 2 (RIPv2) (multiple contexts), EIGRP (multiple contexts), OSPF version
2 (OSPFv2) (multiple processes), Intermediate System-to-Intermediate System (IS-IS)
(multiple contexts), and BGP4 (multiple contexts) as routing protocols that can be used per-
VRF to exchange customer routing information between CE routers and PE routers. The VRF
interfaces on PE routers can be either logical or physical, but each interface can be assigned to
only one VRF.
The customer routes from VRF tables are exported as VPNv4 routes into MP-BGP and
propagated to other PE routers. The MP-BGP sessions between the PE routers are IBGP
sessions and are subject to the IBGP split-horizon rules. Either a full mesh of MP-IBGP
sessions is required between PE routers, or route reflectors need to be used.

1-46 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 -~--_.
An MP-BGP update contains these elements:
VPNv4 address
Extended communities (for example, route targets)

Label used for VPN packet forwarding

The receiving PE router imports the incoming

VPNv4 routes into the appropriate VRF based
on route targets attached to the routes. Import
route target attribute is matched.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

An MP-BGP update exchange between PE routers contains these elements:

VPNv4 address
Extended BGP communities (for example, RTs are required)
Label used for VPN packet forwarding
Mandatory BGP attributes (for example, AS path). Optionally, the MP-BGP update can
contain any other BGP attribute, such as local preference, multi-exit discriminator (MED),
or standard BGP community.
The PE routers receiving MP-BGP updates import the incoming VPNv4 routes into their VRFs
based on RTs attached to the incoming routes and based on import RTs configured in the
VRFs. The VPNv4 routes installed in the VRFs are converted to IPv4 routes and then
propagated to the CE routers.
The RTs that are attached to a route and the import RTs that are configured in the VRF direct
the propagation of the routes to the CE router. Incoming VPNv4 routes are imported into VRFs
on the receiving PE router only if at least one RT attached to the route matches at least one
import RT that is configured in the VRF.

2012 Cisco Systems, Inc. VPN Technologies 1-47

MPLS VPN Forwarding Mechanisms
This topic describes VPN label propagation between PE routers and the MPLS VPN end-to-end
forwarding mechanism.

Approach 1: The PE routers will label the VPN packets with an LOP label
for the egress PE router, and forward the labeled packets across the MPLS
backbone.
Results:
The P routers perform the label switching, and the packet reaches the egress
PE router.
Because the egress PE router does not know which VRF to use for packet
switching, the packet is dropped.

CE Router CE Router
C2012Ciscoandioritsafflllllte!l.A1I~ghta",served.

A simple MPLS-oriented approach to MPLS VPN packet forwarding across the MPLS VPN
backbone would be to label the customer packet with the label assigned by Label Distribution
Protocol (LDP) for the egress PE router. The core routers consequently would never see the
customer IP packet; instead, the core routers would see just a labeled packet targeted toward the
egress PE router. The core routers would perform simple label-switching operations, eventually
delivering the customer packet to the egress PE router. Unfortunately, the customer IP packet
would contain no VPN or VRF information that could be used to perform VRF lookup on the
egress PE router. The egress PE router would not know which VRF to use for packet lookup
and would need to drop the packet.

1-48 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Approach 2: The PE routers will label the VPN packets with a label stack,
using the LOP label for the egress PE router as the top label, and the VPN
label assigned by the egress PE router as the second label in the stack.
Results:
The P routers perform label switching using the top label, and the packet
reaches the egress PE router. The top label is removed .
The egress PE router performs a lookup on the VPN label and forwards the
packet toward the CE router.

CE Router CE Router
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

An MPLS label stack can be used to tell the egress PE router what to do with the VPN packet.
When using the label stack, the ingress PE router labels the incoming IP packet with two labels.
The top label in the stack is the LDP label for the egress PE router; this label guarantees that the
packet will traverse the MPLS VPN backbone and arrive at the egress PE router. The second
label in the stack is assigned by the egress PE router, and tells how to forward the incoming
VPN packet. The second label could point directly toward an outgoing interface, in which case
the egress PE router would perform label lookup only on the VPN packet. The second label
could also point to a VRF, in which case the egress PE router would first perform a label
lookup to find the target VRF, and then perform an IP lookup within the VRF.
Both methods of implementing second labels are used in Cisco lOS, lOS XE, and lOS XR
Software. The second label in the stack points toward an outgoing interface whenever the CE
router is the next hop of the VPN route. The second label in the stack points to the VRF table
for aggregate VPN routes, VPN routes pointing to a null interface, and routes for directly
connected VPN interfaces.
The two-level MPLS label stack satisfies these MPLS VPN forwarding requirements:
The P routers perform label switching on the LDP-assigned label toward the egress PE
router.
The egress PE router performs label switching on the second label (which it has previously
assigned) and either forwards the IP packet toward the CE router or performs another IP
lookup in the VRF that is pointed to by the second label in the stack.

2012 Cisco Systems, Inc. VPN Technologies 1-49

 PHP on the LOP label can be performed on the last P router.
The egress PE router performs label lookup only on the
VPN label, resulting in faster and simpler label lookup.
IP lookup is performed only once-in the ingress PE router.

CE Router CE Router
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Penultimate hop popping (PHP) is the removal of the top label in the stack on the hop prior to
the egress router. PHP can be performed in frame-based MPLS networks. In these networks,
the last P router in the label-switched path (LSP) tunnel pops the LDP label (as previously
requested by the egress PE router through LDP), and the PE router receives a labeled packet
that contains only the VPN label. In most cases, a single label lookup performed on that packet
in the egress PE router is enough to forward the packet toward the CE router. The full IP
lookup through the forwarding information base (FIB) is performed only once, in the ingress
PE router, even without PHP.

1-50 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 stack from the egress PE router?
-
Question: How will the ingress PE router get the second label in the label

Answer: Labels are propagated in MP-BGP VPNv4 routing updates.

Step 2: The VPN label is advertised to all

other PE routers (participating in the
VPNj in an MP-BGP update.

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

The preceding figures showed that an MPLS label stack with the second label is required for
proper MPLS VPN operation. This label was allocated by the egress PE router. This label needs
to be propagated from the egress PE router to the ingress PE routers to enable proper packet
forwarding. MP-BGP was chosen as the propagation mechanism. Every MP-BGP update thus
carries a label assigned by the egress PE router together with the 96-bit VPNv4 prefix.
These steps describe the label propagation between PE routers:
Step 1 The egress PE router assigns a label to every VPN route received from the attached
CE routers and to every summary route summarized inside the PE router. This label
is then used as the second label in the MPLS label stack by the ingress PE routers
when labeling VPN packets. In the figure, the VPN label 38 for destination
192.168.10.0 is assigned by the egress PE router. The VPN labels that are assigned
locally by the PE router can be inspected with the Cisco lOS, lOS XE, and lOS XR
show mpls forwarding vrf vrf-name command.
Step 2 The VPN labels that are assigned by the egress PE routers are advertised to all other
PE routers (participating in the VPN) together with the VPNv4 prefix in MP-BGP
updates. The labels can be inspected with the Cisco lOS and lOS XE show ip bgp
vpnv4 all labels command or the Cisco lOS XR show bgp vpnv4 unicast labels
command on the ingress PE router. The routes that have an input label but no output
label are the routes received from the CE routers (and the input label was assigned
by the local PE router). The routes with an output label but no input label are the
routes received from the other PE routers (and the output label was assigned by the
remote PE router).

2012 Cisco Systems, Inc. VPN Technologies 1-51

 Step 3 The ingress PE router has two labels associated with a remote VPN route: a label for
the next hop assigned by the next-hop P router via LDP-and taken from the local
label information base (LIB)-and also the label assigned by the remote PE router
and propagated via the MP-BGP update. Both labels are combined in a label stack
and installed in the VRF table. The label stack in the VRF table can be inspected
using the Cisco lOS and lOS XE show ip cef vrf vrf-name detail command or the
Cisco lOS XR show cef vrf vrf-name detail command. The tags imposed values in
the output display the MPLS label stack. The first label in the MPLS label stack is
the LDP label forwarded toward the egress PE router, and the second label is the
VPN label advertised by the egress PE router.

The VPN label must be assigned by the BGP next hop.

The BGP next hop should not be changed in the MP-IBGP update
propagation.
- Do not use the next-hap-self command on confederation boundaries.
The PE router must be the BGP next hop.
- Use the next-hap-self command on the PE router.
The label must be reoriginated if the next hop is changed.
- A new label is assigned every time that the MP-BGP update crosses the AS
boundary where the next hop is changed.

MPLS VPN packet forwarding works correctly only if the router specified as the BGP next hop
in the incoming BGP update is the same PE router that assigned the second label in the label
stack. Here are three scenarios that can cause the BGP next hop to be different from the IP
address of the PE router assigning the VPN label:
If the customer route is received from the CE router via an External Border Gateway
Protocol (EBGP) session, the next hop ofthe VPNv4 route is still the IP address of the CE
router (the BGP next hop of an outgoing IBGP update is always identical to the BGP next
hop of the incoming EBGP update). You must configure the next-hop-self command on
the MP-BGP sessions between PE routers to make sure that the BGP next hop of the
VPNv4 route is always the IP address of the PE router, regardless of the routing protocol
used between the PE router and the CE router.
The BGP next hop should not change inside an AS.
The BGP next hop is always changed on an EBGP session. If the MPLS VPN network
spans multiple public autonomous systems, special provisions must be made in the AS
boundary routers to reoriginate the VPN label at the same time that the BGP next hop is
changed.

1-52 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 The VPN label of the BGP route is understood by only the egress PE router.
An end-to-end LSP tunnel is required between the ingress and egress PE routers.
BGP next-hop addresses must be IGP routes.
- LDP labels will be assigned to addresses in the global routing table.
- LDP labels are not assigned to BGP routes (BGP routes receive VPN labels).
BGP next hops announced in IGP must not be summarized in the core network.
- Summarization breaks the LSP tunnel.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

For successful propagation of MPLS VPN packets across an MPLS backbone, there must be an
unbroken LSP tunnel between PE routers. This is because the second label in the stack is
recognized only by the egress PE router that has originated it, and will not be understood by
any other router if it ever becomes exposed.
Here are two scenarios that could cause the LSP tunnel between PE routers to break:
If the IP address ofthe PE router is announced as a BGP route, it will have no
corresponding LDP label, and the label stack will not be built correctly. The IP address of
the PE router must be announced in the global routing table.
If the P routers perform summarization of the address range within which the IP address of
the egress PE router lies, the LSP tunnel will be disrupted at the summarization point.
In the figure, the P router summarizes the loopback address ofthe egress PE router. The LSP
tunnel is broken at a summarization point, so the summarizing router needs to perform full IP
lookup. In an MPLS network, the P router would request PHP for the summary route, and the
upstream P router (or a PE router) would remove the LDP label, exposing the VPN label to the
P router. Because the VPN label is assigned not by the P router but by the egress PE router, the
label will not be understood by the P router and the VPN packet will be dropped or misrouted.

2012 Cisco Systems, Inc. VPN Technologies 1-53

Summary
This topic summarizes the primary points that were discussed in this lesson.

5 g
The most scalable method of exchanging customer routes across a
provider network is the use of an MP-BGP between PE routers.
- RDs transform non-unique 32-bit addresses into 96-bit unique addresses.
- RTs are used to identifyVPN membership in overlapping topologies.
In MPLS VPNs:
- CE routers run standard routing protocols to the PE routers.
- PE routers provide the VPN routing and services via MP-BGP.
- P routers do not participate in VPN routing, and only provide core IGP
backbone routing to the PE routers.
PE routers forward packets across the MPLS VPN backbone using label
stacking.

C2012Ci'"""8rd'oritstrffili*'".A1lrigl'ts......,'......

1-54 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Summary
This topic summarizes the primary points that were discussed in this module.

3
VPNs replace dedicated links with virtual point-to-point links on common
infrastructure, reducing operating costs for customers.
Label stacking is used in forwarding packets across the MPLS VPN
backbone.

02012CilCO.ndlor~.afl'ili_.Allrighlol . . .rv"'.

The two major VPN design options---overlay VPN and peer-to-peer VPN-have many benefits
and drawbacks. The VPN topology categories and architectural components help determine the
method for forwarding packets in a Multiprotocol Label Switching (MPLS) VPN environment.

2012 Cisco Systems, Inc. VPN Technologies 1-55

1-56 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Self-Check
Use the questions here to review what you learned in this module. The correct answers and
solutions are found in the Module Self-Check Answer Key.
QI) Which three options are advantages ofVPNs? (Choose three.) (Source: Introducing
VPNs)
A) cost savings
B) scalability
C) improved security
D) complexity
E) simplified routing
Q2) Which two options are Layer 2 overlay VPN technologies? (Choose two.) (Source:
Introducing VPNs)
A) Frame Relay
B) GRE
C) SSL VPN
D) ATM
Q3) Which two options are Layer 3 overlay VPN technologies? (Choose two.) (Source:
Introducing VPNs)
A) Frame Relay
B) DMVPN
C) IPsec
D) ATM
Q4) Which of these is a peer-to-peer VPN technology? (Source: Introducing VPNs)
A) DMVPN
B) GETVPN
C) IPsec
D) L2TPv3
Q5) Which routers are MPLS VPNs aware of? (Source: Introducing MPLS VPNs)

Q6) Which protocol is used to transport customer routes directly between PE routers?
(Source: Introducing MPLS VPNs)
A) RIP
B) VPN
C) BGP
D) OSPF
Q7) In which two ways do MPLS VPNs support overlapping customer address spaces?
(Choose two.) (Source Introducing MPLS VPNs)
A) by implementing unique RDs for each customer
B) by implementing unique RTs for each customer
C) by implementing different LSPs for each customer
D) by implementing virtual routing spaces for each customer

2012 Cisco Systems, Inc. VPN Technologies 1-57

 Q8) Why do MPLS VPNs implement route targets? (Source: Introducing MPLS VPNs)
A) to identify different customer VPNs
B) to allow a site to participate in more than one VPN
C) to convert a customer address to an MP-BGP address
D) to convert a non-unique IP address into a unique VPNv4 address

Q9) Which type of routers exchange VPNv4 routes? (Source: Introducing MPLS VPNs)
A) P
B) CE
C) PE
D) core

Q 10) Which protocol would a PE router use to support an existing Internet routing scheme?
(Source: Introducing MPLS VPNs)
A) IS-IS
B) EIGRP
C) BGP4
D) BGPVPNv4

Q 11) What is the effect of an MPLS VPN on CE routers? (Source: Introducing MPLS VPNs)
A) The CE routers must support BGP.
B) The CE routers must run a link-state protocol.
C) The CE routers can run any standard IP routing protocol.
D) The IGP of the CE routers must be upgraded to a VPN-aware IGP.

Q 12) Why would IPv4 routing be enabled on the PE router? (Source: Introducing MPLS
VPNs)
A) to support the MPLS VPN route update
B) to support the MPLS VPN route target exports
C) to support an existing Internet routing scheme
D) to support the transport of MP-BGP extended communities

Q 13) Which two types of routes would an MPLS VPN install into the VRF? (Choose two.)
(Source: Introducing MPLS VPNs)
A) those routes received via an IPv4 update
B) those routes received via a VPNv4 update
C) those routes received via the core IGP update
D) those routes received via the customer IGP update

Q14) Which protocol is used to transport VPN labels between PE routers? (Source:
Introducing MPLS VPNs)
A) LDP
B) RSVP
C) MP-BGP
D) the core IGP

Q15) How can P routers forward VPN packets ifthey do not have VPN routes? (Source:
Introducing MPLS VPNs)
A) They forward based on the LSP label.
B) They forward based on the VPN label.
C) They forward based on the MP-BGP next hop.
D) They forward based on a routing table lookup of the IP address.

1-58 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Q16) Which router assigns the VPN label? (Source: Introducing MPLS VPNs)
A) P
B) egress CE
C) egress PE
D) ingress CE
E) ingress PE
QI 7) What is used to identify the label that will be used to transport the VPN packet to the
egress router? (Source: Introducing MPLS VPNs)
A) the IGP least-cost path
B) the EBGP next-hop address
C) the MP-IBGP next-hop address
D) the VPN label entry in the LFIB

2012 Cisco Systems, Inc. VPN Technologies 1-59

Module Self-Check Answer Key
Ql) A,B,C
Q2) A,D
Q3) B,C
Q4) B
Q5) PE routers
Q6) C
Q7) A,D
Q8) B
Q9) C
QlO) C
Qll) C
Q12) C
Q13) B,D
Q14) C
Q15) A
Q16) C
Q17) C

1-60 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Module 21

MPLS Layer 3 VPNs

Overview
This module covers Multiprotocol Label Switching (MPLS) VPN implementation on Cisco
lOS Software platforms. The module describes the concepts of virtual routing and forwarding
(VRF) tables, the interaction between customer-to-provider routing protocols, and
Multiprotocol Border Gateway Protocol (MP-BGP) in the service provider backbone. It also
describes advanced MPLS VPN-specific routing protocol features. The module continues with
a description of the MPLS VPN monitoring and debugging commands that are available on
Cisco lOS platforms and describes troubleshooting, including failure scenarios, symptoms, and
remedial action. At the end, this module describes IPv6 service provider deployment strategies.

Module Objectives
Upon completing this module, you will be able to configure, monitor, and troubleshoot VPN
operations and identify IPv6 strategies in service provider environments. You will be able to
meet these objectives:
Configure VRF tables and MP-BGP sessions between PE routers
Configure small-scale routing protocols (static, RIP, and EIGRP) between CE and PE
routers
Configure OSPF and BGP as the routing protocol between CE and PE routers and explain
how to troubleshoot MPLS VPN operations
Describe various methods that are used to deploy IPv6 over MPLS
2-2 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 11

Implementing MPLS Layer 3

VPN Backbones
Overview
This lesson introduces the virtual routing and forwarding (VRF) table, the major data structure
that is associated with Multiprotocol Label Switching (MPLS) VPN implementation on Cisco
lOS, lOS XE, and lOS XR platforms. The lesson describes the MPLS VPN attributes that are
associated with a VRF instance and explains the need for routing protocol contexts and the
interaction of routing protocol contexts, VRFs, and Multiprotocol Border Gateway Protocol
(MP-BGP).
Having a clear understanding of how information is exchanged using VRFs and routing
protocol contexts will make it easier to configure VRFs in your network.
This lesson also explains how to configure VRF tables, listing the configuration tasks, syntax,
and definitions of commands that are used to create VRFs. It provides an example of a VPN
configuration.
It is important to know how to configure and apply a VRF table onto a routing interface. It is
essential to understand the command syntax for the configurations that you want to deploy in
your network. This lesson provides you with the information that will enable you to succeed at
such tasks.

Objectives
Upon completing this lesson, you will be able to describe the VRF table and other MPLS VPN
attributes that are associated with a VRF instance. You will be able to meet these obj ectives:
Describe VRF
Enable VRF
Enable MP-BGP
Virtual Routing and Forwarding
This topic describes the VRF table and describes the other MPLS VPN attributes that are
associated with a VRF instance.

Customers connect to service provider via IP

Service provider uses MPLS to forward packets between edge routers
Service provider enables any-to-any connectivity between sites
belonging to the same VPN
Service provider uses virtual routers to isolate customer routing
information
Customers can use any addressing inside their VPN

IP
+

VPNA

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

The main characteristic of Layer 3 MPLS VPNs is that customers connect to the service
provider via IP. They need to establish IP routing (static or dynamic) to exchange routing
information between customer sites that belong to the same VPN. Because different customers
might use the same private IP address ranges, the service provider cannot perform normal IP
forwarding. Instead, service providers must use MPLS to ensure isolation in the data plane
between packets belonging to different customers but potentially having the same IP addresses.
Virtual routers (VRF instances) are used on service provider routers to isolate customer routing
information. MPLS seamlessly provides any-to-any connectivity between sites that belong to
the same VPN.

2-4 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 A VRF is the routing and forwarding instance for a set of sites with
identical connectivity requirements.
Data structures associated with a VRF are as follows:
- IP routing table
- Cisco Express Forwarding table
- Set of rules and routing protocol parameters (routing protocol contexts)
- List of interfaces that use the VRF
Other information associated with a VRF is as follows:
- Route distinguisher
- Set of import and export route targets

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts......".......

The major data structure that is associated with MPLS VPN implementation on Cisco lOS
platforms is the VRF table. This data structure encompasses an IP routing table that is identical
in function to the following:
The global IP routing table in Cisco lOS Software
A Cisco Express Forwarding table that is identical in function to the global Cisco Express
Forwarding table (forwarding information base [FIB])
Specifications for routing protocols running inside the VRF instance

A VRF is a routing and forwarding instance that you can use for a single VPN site or for many
sites connected to the same provider edge (PE) router, if and only if these sites share exactly the
same connectivity requirements.
Other MPLS VPN attributes that are associated with a VRF table are as follows:
The route distinguisher (RD), which is prepended (for example, RD + IP address) to all
routes that are exported from the VRF into the global VPNv4 (also called VPN IPv4) BGP
table
A set of export route targets (RTs), which are attached to any route that is exported from
the VRF
A set of import RTs, which are used to select VPNv4 routes that are to be imported into the
VRF

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-5

 10.1.1.0/24 There are two backbones with
overlapping addresses.

VPNA

MPLSVPN
Backb=on"'e~-

VPN B CE-B Address Conflict

10.1.1.0/24 RIP is running in both VPNs.

RIP in VPN A has to be different from
RIP in VPN B.
Cisco lOS Software supports only
one RIP process per router.

C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

Traditional Cisco lOS Software can support a number of different routing protocols. In some
cases, even several completely isolated copies of the same routing protocol are supported. For
example, several Open Shortest Path First (OSPF) processes can be used.
It is important to understand that for several important routing protocols, such as Routing
Information Protocol (RIP), Enhanced Interior Gateway Routing Protocol (EIGRP),
Intermediate System-to-Intermediate System (IS-IS), and Border Gateway Protocol (BGP),
Cisco lOS Software supports only a single copy of the protocol running in the router. These
protocols cannot be used directly between PE and customer edge (CE) routers in VPN
environments because each VPN (or, more precisely, each VRF) needs a separate, isolated
copy of the routing protocol to prevent undesired route leakage between VPNs. Furthermore,
VPNs can use overlapping IP address spaces (for example, each VPN could use subnetworks of
network 10.1.1.0/24), which would also lead to routing confusion if all VPNs shared the same
copy of the routing protocol.

2-6 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Routing context =routing protocol run in one VRF
Supported by VPN-aware routing protocols: External BGP (EBGP),
EIGRP, OSPF, RIPv2, IS-IS, static routes
Implemented as several instances of a single routing process
(EIGRP, EBGP, RIPv2, IS-IS) or as several routing processes (OSPF)
Independent per-instance router variables for each instance

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts......".......

"Routing contexts" were introduced in Cisco lOS Software to support the need for separate
isolated copies ofVPN routing protocols. Routing contexts can be implemented as separate
routing processes (in OSPF), similar to a traditional Cisco lOS Software implementation, or as
separate isolated instances of the same routing protocol.
If routing contexts are implemented as instances of the same routing protocol, each instance
contains its own independent routing protocol parameters. Examples would include networks
over which the routing protocol is run, timers, authentication parameters, passive interfaces,
and neighbors. This independence allows the network designer to have maximum flexibility in
implementing routing protocols between PE and CE routers.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-7

 Contains routes that should be available to a particular set of sites
Analogous to standard Cisco lOS Software routing table; supports same
set of mechanisms
VPN interfaces (physical interface, subinterfaces, logical interfaces)
assigned to VRFs:
- Many interfaces per VRF
- Each interface assignable to only one VRF

C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

The routes that are received from VRF routing protocol instances or from dedicated VRF
routing processes are inserted into the IP routing table that is contained within the VRF. This IP
routing table supports exactly the same set of mechanisms as the standard Cisco lOS Software
routing table. These mechanisms include filter mechanisms (distribute lists or prefix lists) and
interprotocol route-selection mechanisms (administrative distances).
The per-VRF forwarding information base (FIB) table is built from the per-VRF routing table.
This table is used to forward all the packets that are received through the interfaces that are
associated with the VRF. Any interface can be associated with a VRF, whether it is a physical
interface, subinterface, or logical interface, as long as it supports Cisco Express Forwarding
switching.
There is no limit to the number of interfaces that can be associated with one VRF (other than
the number of interfaces that are supported by the router). However, each interface can be
assigned to only one VRF because the router needs to uniquely identify the forwarding table to
be used for packets that are received over an interface.

2-8 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 - PE Router
VRF-A Routing Table BGPRouting
Process

Backbone
VRF-B Routing Table Multiprotocol
BGP

CE-BGP-A
Instance for VRF-B

CE-BGP-B

Two VPNs are attached to the same PE router.

Each VPN is represented by a VRF.

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts......".......

This figure and the next fignres illustrate the interactions between VRF instances of routing
processes, VRF routing tables, and the global VPNv4 BGP routing process.
The network contains two VPN customers. Ordinarily, the customer sites would be connected
to a number ofPE routers. This example focuses on only a single PE router, which contains two
VRFs-one for each customer. Each customer is connected to the PE router, which is running
BGP. CE-BGP-A is the CE router for customer A and is associated with VRF-A (VPN-A). CE-
BGP-B is the CE router for customer B and is associated with VRF-B (VPN-B). Both CE
routers are using BGP for exchanging routes with the PE router.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-9

 --
PE Router
VRF-A Routing Table BGPRouting
Process

Backbone
VRF-B Routing Table Multiprotocol
BGP

CE-BGP-A
Instance for VRF-B

CE-BGP-B

BGP-speaking CE routers announce their prefixes to the PE router via BGP.

The instance of the BGP process associated with the VRF of the PE-CE interface collects
the routes and inserts them into the VRF routing table.

C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

The CE routers are BGP neighbors of the PE router. The BGP-speaking CE routers announce
their networks via External Border Gateway Protocol (EBGP) sessions to the PE router. The PE
router associates each BGP neighbor relationship with individual VRFs. The routes that are
received from each VRF routing protocol instance are inserted into the IP routing table that is
contained within that VRF.
A per-VRF forwarding table, a FIB, is built from the per-VRF routing table and is used to
forward all the packets that are received through the interfaces associated with the VRF.

2-10 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 - -- PE Router
VRF-A Routing Table BGPRouting
Process

Backbone
VRF-B Routing Table Multiprotocol
BGP

CE-BGP-A

. Instance for VRF-B

CE-BGP-B

The route distinguishers are prepended during the route export to the BGP routes from the
VRF instance of the BGP process to convert them into VPNv4 prefixes. Route targets are
attached to these prefixes.
VPNv4 prefixes are propagated to other PE routers.

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts......".......

This figure illustrates the interactions between VRF instances of routing processes, VRF
routing tables, and the global VPNv4 BGP routing process.

Example: BGP Route Propagation-Outbound

The BGP routes that are received from BGP-speaking CE routers are copied into the MP-BGP
table for further propagation to other PE routers. This is the export process.
The IP prefixes are prepended with the RD, and the set ofRTs (extended BGP communities)
that are configured as export RTs for the VRF is attached to the resulting VPNv4 route.

Note There are not separate per-VRF BGP and global MP-BGP tables in Cisco lOS Software.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-11

 PE Router

CE-BGP-A

CE-BGP-B

VPNv4 prefixes are received from other PE routers.

The VPNv4 prefixes are inserted into the proper VRF routing tables based
on their route targets and the import route targets configured in VRFs.
The route distinguisher is removed during this process.
C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

As other PE routers start originating VPNv4 routes, the MP-BGP process in the PE router
receives the routes. The routes are filtered, based on the RT attributes attached to them, and are
inserted into the proper per-VRF IP routing tables, based on the import RTs that are configured
for individual VRFs. The RD that was prepended by the originating PE router is removed
before the route is inserted into the per-VRF IP routing table.

PE Router

Routes are received from backbone MP-BGP and imported into a VRF.
IPv4 routes are forwarded to EBGP CE neighbors attached to
thatVRF.

C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

The Multiprotocol Internal Border Gateway Protocol (MP-IBGP) VPNv4 routes that are
received from other PE routers and selected by the import RTs of a VRF are automatically
propagated as 32-bit IPv4 routes to all BGP-speaking CE neighbors of the PE router.

2-12 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 PE Router

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts......".......

In this example, you see an incoming routing update for network 172.16.10.0/24 with RD
1: 100. The routing update has also defined RT 1: 100.
Because VRF-A has defined import RT 1:100, a routing update is inserted into the VRF-A
routing table. A routing update is also sent to the CE-BGP-A router.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-13

 PE Router

BGPRouting
Process

Backbone
Multiprotocol
BGP

RIP-speaking CE routers announce their prefixes to the PE router via RIP.

The instance of the RIP process associated with the VRF of the PE-CE interface
collects the routes and inserts them into the VRF routing table.

C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

This example describes the outbound non-BGP route propagation process in an MPLS VPN
implementation. The example describes RIP-speaking CE routers, but the process would be
similar for other non-BGP protocols.
RIP-speaking CE routers identify the correct instance of RIP on the PE router when an inbound
PE interface is associated with a VRF. This association allows CE routers to announce their
networks to the appropriate per-VRF routing table.

2-14 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 PE Router

BGPRouting
Process

Backbone
Multiprotocol
BGP

The RIP routes entered in the VRF routing table are redistributed into BGP
for further propagation into the MPLS VPN backbone.
Redistribution between RIP and BGP has to be configured for proper
MPLS VPN operation.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

MP-BGP is used in the MPLS VPN backbone to carry VPN routes (prefixed with the RD) as
96-bit VPNv4 routes between the PE routers. The backbone BGP process looks exactly like a
standard Internal Border Gateway Protocol (IBGP) setup from the perspective of the VRF. The
per-VRF RIP routes therefore must be redistributed into the per-VRF instance of the BGP
process to allow them to be propagated through the backbone MP-BGP process to other PE
routers.

Caution Failure to redistribute non-BGP routes into the per-VRF instance of BGP is one of the most
common MPLS VPN configuration errors.

If there is an overlap between an inbound RIP update and an inbound EBGP update, the
standard route-selection mechanism (administrative distance) is used in the per-VRF IP routing
table, and the EBGP route takes precedence over the RIP route. EBGP precedence results from
the fact that the administrative distance ofEBGP routes (20) is better than the administrative
distance of RIP routes (120).

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-15

 PE Router

RIP Routing Process
BGPRouting
Instance for VRF-A Process

CE-RIP-A Backbone
Instance forVRF-B Multiprotocol
BGP
CE-RIP-B

The RIP routes entered in the VRF routing table are redistributed into BGP
for further propagation into the MPLS VPN backbone.
Redistribution between RIP and BGP has to be configured for proper
MPLS VPN operation.

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

The MP-IBGP routes, although they are inserted in the per-VRF IP routing table, are not
propagated to RIP-speaking CE routers automatically. To propagate these MP-IBGP routes to
the RIP-speaking CE routers, you must manually configure redistribution between the per-VRF
instance ofBGP and the per-VRF instance of RIP.

PE Router

BGPRouUng
Process

Backbone
Multiprotocol
BGP

Routes redistributed from BGP into a VRF instance of RIP are sent to RIP-speaking CE
routers.

C2012Ci'"""8rd'oritstrffili*'".A1lrigl'ts......,'......

When the IBGP routes from the per-VRF IP routing table are successfully redistributed into the
per-VRF instance of the RIP process, the RIP process announces these routes to RIP-speaking
CE routers, thus achieving transparent end-to-end connectivity between the CE routers.

2-16 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 PE Router

BGPRouting
Process

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

In this example, you can see a routing update for network 172.16.10.0/24. A route is inserted in
the VRF-A routing table. The routing update is processed by the VRF-A instance of the BGP
routing process and redistributed to the VRF-A instance of the RIP routing process. The routing
update is then sent to the CE-RIP-A router by the RIP routing process.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-17

Enabling VRF
This topic describes how to enable VRF.

Create a VRF table. Router (config) # ip vrf vrf-na.me

Assign an RO to the VRF. RoutQr(config-vrf)# rd route-distinguish~r

Router(config-vrf)# route-target export RT

Specify export and import route
targets. Router(config-vrf)# route-target import RT

Configure a VPN 10 (optional). RoutQr(config-vrf)# vpn id oui:vpn-ind~x

Assign interfaces to VRFs. Router(config-if)# ip vrf forwarding vr-name

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts"'""'v.....

Configuring a VRF table and starting deployment of an MPLS VPN service for a customer on
Cisco lOS and lOS XE platform consists of these four mandatory steps:
1. Create a new VRF table.

2. Assign a unique RD to the VRF.

Note You must assign a unique RD to every VRF created in a PE router. The same RD might be
used in multiple PE routers, based on customer connectivity requirements. The same RD
should be used on all PE routers for simple VPN service.

3. Specify import and export RTs for the VRF.

Note Import and export RTs should be equal to the RD for simple VPN service.

4. Assign interfaces to the VRF.

2-18 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEOGE) v1.0 2012 Cisco Systems, Inc.
 Create a VRF table. RPjOjRPOjCPUO:router(config)# vrf vr-name

Enter VRF address

family configuration RPjOjRPOjCPUO:router(config-vrf)# 5ddress-family ipv4 unicast
mode for the IPv4
address family.
Specify import route RPjOjRPOjCPUD:router(config-vrf-af) # import route-target [as-

targets. number:nn I ip-address:nn]

Specify export route RPjOjRPOjCPUD:router(config-vrf-af) # export routlil-target [as-

targets. numb~r:nn I ip-addr~ss:nn]

Assign interfaces RPjOjRPOjCPUD:router(config-if)# vrf vrf-name

to VRFs.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

Configuring a VRF table on Cisco lOS XR devices is somewhat different from using Cisco lOS
and lOS XE Software. Basic configuration consists of these four mandatory steps:
Step 1 Create a new VRF table.
Step 2 Enter the IPv4 unicast address family configuration.
Step 3 Specify import and export RTs for the VRF.
Step 4 Assign interfaces to the VRF.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-19

 -"31II IF l:iiiiII.
Router (config)#

Cisco lOS and

10SXE
lip vrf vrf-name

RP/O/RPO/CPUO:router(config)#

Cisco lOS XR Ivrf vrf-name

This command creates a new VRF or enters configuration of an

existing VRF.
VRF names are case-sensitive.
VRF names have only local significance.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts"'""'v.....

To configure a VRF routing table on the Cisco lOS and lOS XE platforms, use the ip vrf
command in global configuration mode. To remove a VRF routing table, use the no form of
this command.
ip vrf vrj-name
no ip vrf vrj-name

To configure a VRF routing table on the Cisco lOS XR platform, use the vrf command in
global configuration mode. To remove a VRF routing table, use the no form of this command.
vrf vrf-name
no vrf vrfname

No VRFs are defined by default. No import or export lists are associated with a VRF. No route
maps are associated with a VRF.

2-20 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 This command assigns a route distinguisherto a VRF.
A VRF is not operational unless you configure an RD.
You can use the ASN:nn or A.B.C.D:nn format for RD.
Each VRF in a PE router must have a unique RD.

Cisco lOS and lOS XE configuration

RD is configured under VRF configuration area

Router(config)#ip vrf vrf-name

Router (config-vrf)#rd route-distinguisher

Cisco lOS XR configuration

RD is configured under BGP configuration area

Router(config)#router bgp AS
Router(config-bgp)#vrf vrf-name
Router(config-bgp-vrfJ#rd route-distinguisher

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

To create routing and forwarding tables for a VRF on the Cisco lOS and lOS XE platforms, use
the rd command in VRF configuration submode: rd route-distinguisher. The route-
distinguisher parameter adds an 8-byte value to an IPv4 prefix to create a VPNv4 prefix.
The RD can be specified in one of these two formats:
l6-bit autonomous system (AS) number followed by a 32-bit decimal number (ASN:nn)
32-bit IP address followed by a 16-bit decimal number (A.B.C.D:nn)

There is no default for this command. An RD must be configured for a VRF table to be
functional.
To create routing and forwarding tables for a VRF on a Cisco lOS XR operating system, use
the rd command in the BGP configuration area in VRF configuration submode.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-21

 1IIIIF
RP/O/RPO/CPUO: router (config-vrf) #

Iaddress-family
-
ipv4 unicast

Cisco lOS XR Software only

This command allows you to enterVRF address family configuration
mode forthe IPv4 address family

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

To create routing and forwarding tables for a VRF on the Cisco lOS XR platform, you must
first enter VRF address family configuration submode using the address-family ipv4 unicast
command. Address families are used within VRF configuration mode to control import and
export policies.

2-22 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config-vrf) #

Iroute-target export RT
Specifies an RT to be attached to every route exported from this
VRF to Multiprotocol Border Gateway Protocol
Allows specification of many export RTs-ali to be attached to
every exported route
Router (config-vrf) #

Iroute-target import RT
Specifies an RT to be used as an import filter. (Only routes
matching the RT are imported into the VRF.)
Allows specification of many import RTs. (Any route where at least
one RT attached to the route matches any import RT is imported
into the VRF.)
Because of implementation issues, in Cisco lOS Release 12.4(T) and earlier, at least
one export route target must also be an import route target of the same VRF.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

To create an RT extended community for a VRF on the Cisco lOS and lOS XE platforms, use
the route-target command in VRF submode. To disable the configuration of an RT community
option, use the no form of this command.
route-target {import I export I both} route-target-ext-community
no route-target {import I export I both} route-target-ext-community
This table describes the parameters for the route-target command.

Syntax Description

Parameter Description

import Imports routing information from the target VPN extended

community

export Exports routing information to the target VPN extended

community

both Sets the value to be used by both the import and export process
to the value that is indicated in the route-target-ext-community
field

route-target-ext-community Adds the route target extended community attributes to the VRF
list of import, export, or both (import and export) route target
extended communities

Similar to RDs, RTs can be specified in one of these two formats:

16-bit AS number followed by a 32-bit decimal number (ASN:nn)
32-bit lP address followed by a 16-bit decimal number (A.B.C.D:nn)

A VRF has no RT extended community attributes associated with it until they are specified by
the route-target command.
Whenever an RT is both an import RT and an export RT for a VRF, you can use the route-
target both command to simplify the configuration.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-23

 RP/O/RPO/CPUO:router(config-vrf-af)#

export route-target [as-number:nn I ip-address:nnJ

Allows specification of many export RTs-ali to be attached to every
exported route

RP/O/RPO/CPUO:router(config-vrf-af)#

import route-target [as-number:nn I ip-address:nnJ

Allows specification of many import RTs. (Any route where at least one RT
attached to the route matches any import RT is imported into the VRF.)

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

The Cisco lOS XR export route-target command associates the local VPN with an RT. When
the route is advertised to other PE routers, the export RT is sent along with the route as an
extended community.
The import route-target command allows exported VPN routes to be imported into the VPN if
one of the RTs of the exported route matches one of the local VPN import RTs.

2-24 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 A VPN identifier (VPN 10) allows you to identifyVPNs by an 10 number.
- Not used to control distribution of routing information
- Not used to associate IP addresses with VPN IDs in routing updates
- Is stored on the VRF structure for a VPN
Has the following elements:
- aUI (three-octet hexadecimal number)
- A VPN index (four-octet hexadecimal number identifying the VPN within the
company)
Configure all PE routers that belong to the same VPN with the same
VPNIO.
Make the VPN 10 unique to the service provider network.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

The MPLS VPN ID is an optional feature that allows you to identify VPNs by a VPN
identification number. The MPLS VPN ID feature is not used to control the distribution of
routing information or to associate IP addresses with MPLS VPN ID numbers in routing
updates.
You can configure all the PE routers that belong to the same VPN with the same VPN ID.
Make sure that the VPN ID is unique to the service provider network.
The VPN ID is stored in the corresponding VRF structure for the VPN. To ensure that the VPN
has a consistent VPN ID, assign the same VPN ID to all the routers in the service provider
network that service that VPN.
Each VPN ID that is defined by RFC 2685 consists ofthese elements:
An Organizationally Unique Identifier (OUI), a three-octet hexadecimal number that is
assigned by the IEEE
A VPN index, a four-octet hexadecimal number that identifies the VPN within the
company
A VPN ID is useful for remote access applications, such as RADIUS and DHCP, which can use
the MPLS VPN ID to identify a VPN. RADIUS can use the VPN ID to assign dial-in users to
the proper VPN, based on the authentication information of each user.

Note You can use a VRF name (a unique ASCII string) to reference a specific VPN that is
configured in the router. Alternatively, you can use a VPN ID to identify a particular VPN that
is configured in the router. The VPN name is not affected by the VPN ID configuration.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-25

 Router (config) #

liP vrf vrf-name

Creates a VRF routing table and a Cisco Express FOlWarding table
and enters VRF configuration mode

Router (config-vrf) #

Ivpn id oui:vpn-index
Assigns the VPN 10 to the VRF

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

To assign a VPN ID to a VRF, use the vpn id command in VRF configuration submode. To
disable the configuration of an RT community option, use the no form of this command.
vpn id oui:vpn-index
no vpn id oui:vpn-index
This table describes the parameters for the vpn id command.

Syntax Description

Parameter Description

aui Identifies the aUI, which is restricted to three octets and is

followed by a colon

vpn-index Identifies the VPN within the company and is restricted to four
octets

Each VRF configured in a PE router can have a VPN ID configured. Configure all the PE
routers that belong to the same VPN with the same VPN ID. The VPN ID should be unique to
the service provider network.

2-26 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config-if)#

Cisco lOS and lip vrf forwarding vrf-name

10SXE
RP/O/RPO/CPUO:router(config-if)#

Cisco lOS XR Ivrf vrf-name I

This command associates an interface with the specified VRF.
The existing IP address is removed from the interface when the interface
is put into the VRF-the IP address must be reconfigured.
Cisco Express Forwarding switching must be enabled on the interface.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

To associate a VRF with an interface or subinterface, use the ip vrf forwarding command on
the Cisco lOS and lOS XE platforms or the vrf command on the Cisco lOS XR platform in
interface configuration mode. To disassociate a VRF, use the no form of this command.

Note You must remove IPv4 and IPv6 addresses from an interface before assigning, removing, or
changing its VRF. If you do not, any attempt to change the VRF on an IP interface is
rejected.

After local interfaces are bound to the VRF, the interfaces appear in the routing display of the
VRF table.

Note When an interface is configured with a particular VRF, its IP address is removed from the
interface and from the global routing table. This action is based on the assumption that the
address is not valid across multiple routing tables and that the address should be
reconfigured after the interface is associated to a VRF.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-27

 MPLS VPN Backbone
CE-A1 CE-A2
lOS AS 64500
and lOS XE lOS XR

vrf Customer A
address-family ipv4 unicast
ip vrf Customer_A import route-target 64500:11
rd 6111: 11 export route-target 64500:11
route-target both 64500:11
vrf Customer B
ip vrf Customer_B address-family ipv4 unicast
rd 6111: 12 import route-target 64500:12
route-target both 64500:12 export route-target 64500:12

interface GigabitEthernetljOjO interface GigabitEthernetOjOjOjO

ip vrf forwarding Customer_A vrf Customer A
ip address 10.1.0.1 255.255.255.252 ipv4 address 10.1.0.1 255.255.255.252

interface GigabitEthernetl/ljO interface GigabitEthernetOjOjOj2

ip vrf forwarding Customer_B vrf Customer B
ip address 10.2.0.1 255.255.255.252 ipv4 address 10.2.0.5 255.255.255.252

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts"'""'v.....

To illustrate the use ofMPLS VPN configuration commands, you can look at a configuration of
the PE routers in a sample network.
The figure shows configuration of the PE routers in a sample network with two VPN
customers. The PE-X router is running Cisco lOS or lOS XE Software. The PE-Y router is
running Cisco lOS XR Software.
The configuration steps that you perform on the PE router are as follows:
Step 1 Configure VRFs for customer A and customer B.
Step 2 Assign RDs and RTs to the VRFs.

Only one RD per customer is used on all PE routers in this MPLS VPN backbone,
because these customers require only simple VPN connectivity. To simplify the
configuration and troubleshooting process, the RTs are made equal to the RDs.
Step 3 Assign PE-CE interfaces to individual VRFs.

2-28 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 1IIIIF-
"VRF-Lite" equals "VRF without the need to run MPLS between the PE
and CE."
VRF-Lite is a feature that enables a service providerto support two or
moreVPNs.
VRF-Lite includes these devices: CE, PE, and routers in a service
provider network.
VRF-Lite interfaces must be Layer 3 interfaces.
Multiple customers can share one CE, and only one physical link is used
between the CE and the PE.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

Multi-VRF Customer Edge (VRF-Lite) provides the ability to configure and maintain more
than one VRF instance within the same CE router.
VRF-Lite uses input interfaces to distinguish routes for different VPNs and forms VRF tables
by associating one or more Layer 3 interfaces with each VRF. Interfaces in a VRF can be either
physical, such as Ethernet ports, or logical, such as VLAN switch virtual interfaces (SVIs).
However, a Layer 3 interface cannot belong to more than one VRF at any time. The VRF-Lite
feature thus allows an operator to support two or more routing domains on a CE router, with
each routing domain having its own set of interfaces and its own set of routing and forwarding
tables.
VRF-Lite includes these devices:
Customer edge devices: CE devices provide customer access to the service provider
network over a data link to one or more PE routers. The CE device advertises the local
routes of the site to the PE router and learns the remote VPN routes from it. A Cisco
Catalyst 4500 Series Switch can be aCE.
Provider edge routers: PE routers exchange routing information with CE devices by using
static routing or a routing protocol such as BGP, RIPv 1, or RIPv2. The PE router is only
required to maintain VPN routes for the VPNs to which it is directly attached, eliminating
the need for the PE router to maintain all of the service provider VPN routes. Each PE
router maintains a VRF for each of its directly connected sites. Multiple interfaces on a PE
router can be associated with a single VRF if all of these sites participate in the same VPN.
Each VPN is mapped to a specified VRF. After learning local VPN routes from CE
devices, a PE router exchanges VPN routing information with other PE routers by using
IBGP.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-29

 1IIIIF-
CE and PE
. . " ~~:~O,Ofi9U",1;OO
fa1/0
~
~.-.~
rd64500:1
~~~~I route-target both 64500:1

fa3/0 fa1/0 ip vrf VPN-2

rd 64500:2
route-target both 64500:2

interface fastethernetljO interface fastethernetljO.10

ip vrf forwarding VPN-l ip vrf forwarding VPN-l
ip address 10.0.1.1 255.255.255.0 ip address 192.168.1.2 255.255.255.0

interface fastethernetl/l interface fastethernetljO.20

ip vrf forwarding VPN-2 ip vrf forwarding VPN-2
ip address 10.0.2.1 255.255.255.0 ip address 192.168.2.2 255.255.255.0

interface fastethernet3jO.10
ip vrf forwarding VPN-l
ip address 192.168.1.1 255.255.255.0

interface fastethernet3jO.20
ip vrf forwarding VPN-2
ip address 192.168.2.1 255.255.255.0

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

With VRF-Lite, multiple customers can share one CE, and only one physical link is used
between the CE and the PE. The shared CE maintains separate VRF tables for each customer
and switches or routes packets for each customer, based on its own routing table. VRF-Lite
extends limited PE functionality to a CE device, giving it the ability to maintain separate VRF
tables to extend the privacy and security of a VPN to the branch office.
To illustrate VRF-Lite configuration, you can look at a configuration of the CE and PE routers
in a sample network. First, VRFs must be configured on both the PE and the CE routers.
Additionally, you must specify the Layer 3 interface to be associated with the VRF and
associate the VRF with the Layer 3 interface.

2-30 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ...=-_E
router ospf 101 vrf VPN-l
network 10.0.1.0 255.255.255.0 area 0
redistribute bgp 64500 subnets
!
router ospf 102 vrf VPN-2
network 10.0.2.0 255.255.255.0 area 0
redistribute bgp 64500 subnets
!
router bgp 64500 router bgp 64500
address-family ipv4 vrf VPN-l address-family ipv4 vrf VPN-l
neighbor 192.168.1.2 remote-as 64500 neighbor 192.168.1.1 remote-as 64500
neighbor 192.168.1.2 activate neighbor 192.168.1.1 activate
redistribute ospf 101
address-family ipv4 vrf VPN-2
address-family ipv4 vrf VPN-2 neighbor 192.168.2.1 remote-as 64500
neighbor 192.168.2.2 remote-as 64500 neighbor 192.168.2.1 activate
neighbor 192.168.2.2 activate
redistribute ospf 102 interface fastethernet1JO .10
ip vrf forwarding VPN-1
interface fastethernet3JO .10 ip address 192.168.1.2 255.255.255.0
ip vrf forwarding VPN-l
ip address 192.168.1.1 255.255.255.0 interface fastethernet1JO. 20
! ip vrf forwardingip VPN-2
interface fastethernet3jO. 20 ip address 192.168.2.2 255.255.255.0
ip vrf forwardingip VPN-2 !
ip address 192.168.2.1 255.255.255.0
!
C2012Ci:OC08rd'oritstrflili*'".A1lrigi'ts......".......

Most routing protocols can be used between the CE and the PE: BGP, OSPF, EIGRP, RIP, and
static routing. However, EBGP is recommended:
BGP does not require more than one algorithm to communicate with a multitude of CEs.
BGP is designed to pass routing information between systems that are run by different
administrations.
BGP makes it easy to pass attributes of the routes to the CEo

Furthermore, when BGP is used as the routing protocol, it can also be used to manage the
MPLS label exchange between the PE and CE devices. By contrast, ifOSPF, EIGRP, RIP, or
static routing is used, Label Distribution Protocol (LDP) must be used to signal labels.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-31

Enabling MP-BGP
MP-BGP is responsible for allocating labels for VPN routes and advertising them to other edge
routers. This topic describes how to enable MP-BGP.

MP-BGP

Layer3 MPLS VPNs are implemented using MP-BGP to exchange VPN

routing information.
MP-BGP is BGP version 4 with extensions to support other protocols
and applications:
- Layer3 MPLS VPNs
- Virtual Private LAN Services (VPLS) using BGP autodiscovery

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts"'""'v.....

VPNs based on MPLS require an additional VPN label to distinguish between potentially
overlapping prefixes belonging to different VPNs. MP-BGP is BGP version 4 with additional
attributes to support the exchange ofVPN prefixes.
Virtual Private LAN Services (VPLS) can also be implemented using the BGP autodiscovery
feature to simplify the management ofVPLS.
The figure shows an end-to-end MP-IBGP session. This figure is a simplified representation of
the BGP capability to propagate VPN routing information between edge label switch routers
(LSRs). In real environments with many more PE routers, a route reflector would be used
between the edge routers, although that addition would not alter the operation significantly.

2-32 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 MP-BGP

MP-BGP must be configured on edge routers only.

Support for MPLS VPNs must be enabled.
Steps required:
- Add address family vpnv4
- Activate neighbor in address family vpnv4
Optional configuration settings
C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

MP-BGP extension is used in the MPLS world to relay VPN information between two edge
routers. The RD is a 64-bit value that is used to mark prefixes and to separate different
customers.
VPN support in BGP is enabled by configuring a VPNv4 address family. This allows MP-BGP
neighbor sessions to be established independently from existing IPv4 BGP sessions. These
VPNv4 adjacencies are used to relay VPN prefixes together with 64-bit extended communities,
where the RT value is stored. The total length of the VPNv4 address is thus 96 bits.
Configuring the VPNv4 address family and activating neighbors in it is the minimum required
configuration. Optionally, fine-tuning can be performed by adjusting the BGP timers.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-33

 .. The BGP process in an MPLS VPN-enabled router performs three
separate tasks:
- Global BGP routes (Internet routing) are exchanged as in a traditional BGP
setup.
- VPNv4 prefixes are exchanged through MP-BGP.
- VPN routes are exchanged with CE routers through per-VRF External Border
Gateway Protocol sessions or through route redistribution .
Address families (routing protocol contexts) are used to configure these
three tasks in the same BGP process.

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

Independently from the MPLS VPN architecture, the PE router can use BGP IPv4 route updates
to receive and propagate Internet routes in situations where the PE routers are also used to
provide Internet connectivity to customers.
The MPLS VPN architecture uses the BGP routing protocol in these two ways:
VPNv4 routes are propagated across an MPLS VPN backbone using MP-BGP between the
PE routers.
BGP can be used as the PE-CE routing protocol to exchange VPN routes between the PE
routers and the CE routers.
All three route-exchange mechanisms take place in one BGP process (because only one BGP
process can be configured per router). The routing protocol contexts (called address families
from the router configuration perspective) are used to configure all three independent route-
exchange mechanisms.

2-34 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config) #
Irouter bgp as-number
Selects global BGP routing process

Router (config-router) #
Iaddress-family vpnv4
Selects configuration ofVPNv4 prefix exchanges under MP-BGP
sessions

Router (config-router) #
Iaddress-family ipv4 vrf vrf-name

Selects configuration ofper-VRF PE-CE EBGP parameters

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

To configure the BGP routing process, use the router bgp command in global configuration
mode. To remove a routing process, use the no form of this command.
router bgp as-number
no router bgp as-number

Use the address-family command in router configuration mode to select the routing context
that you would like to configure:
Internet routing (global IP routing table) is the default address family that you configure
when you start configuring the BGP routing process.
To configure MP-BGP sessions between the PE routers, use the address-family vpnv4
command.
To configure BGP between the PE routers and the CE routers within individual VRF tables,
use the address-family ipv4 vrfvr:f-name command.

To enter address-family submode for configuring routing protocols, such as BGP, RIP, and
static routing, use the address-family command in global configuration mode. To disable
address-family submode for configuring routing protocols, use the no form of this command.
VPNv4 unicast: address-family vpnv4 [unicast]
Configures sessions that carry customer VPNv4 prefixes, each of which has been made
globally unique by adding an 8-byte RD
IPv4 unicast: address-family ipv4 [unicast]
Configures sessions that carry standard IPv4 address prefixes
IPv4 unicast: address-family ipv4 [unicast] vrf vrf-name
Specifies the name of a VPN VRF to associate with submode commands

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-35

 (

RPjOjRPOjCPUO:router(config) #
Irouter bgp as-number

Selects global BGP routing process

RPjOjRPOjCPUO:router(config-bgp) #
Iaddress-family vpnv4 unicast

Configures VPNv4 prefix

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

Similar to Cisco lOS and IOS XE Software, on Cisco IOS XR Software, use the router bgp
command in global configuration mode.
The VPNv4 address family is configured in the BGP section using the address-family vpnv4
unicast command. Afterwards, it will be applied in the neighbor configuration block.

2-36 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 MP-BGP neighbors are configured under the BGP routing process:
- These neighbors need to be activated for each global address family that they
support.
- Per-address-family parameters can be configured for these neighbors .
VRF-specific BGP neighbors are configured under corresponding
address families.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

MPLS VPN architecture defines these two types ofBGP neighbors:

Global BGP neighbors (other PE routers) with which the PE router can exchange multiple
types of routes. (These neighbors are defined in the global BGP definition and need to be
activated only for individual address families.)
Per-VRF BGP neighbors. (These neighbors are the CE routers.)

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-37

 Router (config)#
router bgp as-number
neighbor ip-address remote-as as-number
neighbor ip-address update-source interface-type
interface-number

All MP-BGP neighbors have to be configured under the global BGP

routing configuration.
MP-IBGP sessions have to run between loopback interfaces.

Router (config-router)#

Iaddress-family vpnv4
This command starts configuration of MP-BGP routing for VPNv4 route
exchange.
The parameters that apply only to MP-BGP exchange of VPNv4 routes
between already configured IBGP neighbors are configured under this
address family.

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

The initial commands that are needed to configure an MP-IBGP session between PE routers are
as follows:
The neighbor ip-address remote-as as-number command configures the neighboring PE
router.
The neighbor ip-address update-source interface-type interface-number command
configures the source address that is used for the TCP session carrying BGP updates and
the IP address that is used as the BGP next hop for VPNv4 routes.
The address-family vpnv4 command allows you to enter VPNv4 configuration mode,
where additional VPNv4-specific parameters must be configured on the BGP neighbor.

2-38 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config-router-af) #
IneighbOr ip-address activate
The BGP neighbor defined under BGP router configuration has to
be activated forVPNv4 route exchange.

Router (config-router-af) #
IneighbOr ip-address next-hop-self
The next-hap-self keyword can be configured on the MP-IBGP
session for MPLS VPN configuration if EBGP is being run with a
CE neighbor.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

After you define the remote PE router as a global BGP neighbor, you must activate it for
VPNv4 route exchange. To enable the exchange of information with a BGP neighboring router,
use the neighbor activate command in router configuration mode. The exchange of addresses
with neighbors is enabled by default for the IPv4 address family. For all other address families,
address exchange is disabled by default. You can explicitly activate the default command by
using the appropriate address family submode.
To enable next-hop processing ofBGP updates on the router, use the neighbor next-hop-self
command in router configuration mode. This command is useful in unmeshed networks (such
as Frame Relay or X.25) where BGP neighbors might not have direct access to all other
neighbors on the same IP subnet. If you specify a BGP peer group by using the peer-group-
name argument, all the members of the peer group inherit the characteristic that is configured
with this command. Specifying the command with an IP address overrides the value that is
inherited from the peer group.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-39

 ~-

RP/O/RPO/CPUO:router(config) #
router bgp as-number
neighbor ip-address remote-as as-number
Configures a neighbor and assigns it a remote autonomous system number

RP/O/RPO/CPUO:router(config-bgp-nbr) #
Iaddress-family vpnv4 unicast
Enters address family configuration mode for the VPNv4 address family.

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

To configure an MP-IBGP neighbor on devices running Cisco lOS XR Software, enter BGP
configuration mode using the command router bgp as-number.
To add a new BGP neighbor, use the command neighbor ip-address remote-as as-number.
In each neighbor configuration area, enable the neighbor for the specific address family.

2-40 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config-router-af) #
neighbor ip-address send-community [standard I extended
I both]

This command with the extended option is enabled by default by Cisco

lOS Software after the BGP neighbor has been activated for VPNv4
route exchange.
The command can be used to enable propagation of standard BGP
communities attached to VPNv4 prefixes.
Usage guidelines:
- Extended BGP communities attached to VPNv4 prefixes have to be
exchanged between MP-BGP neighbors for proper MPLS VPN
operation.
- To propagate standard BGP communities between MP-BGP
neighbors, use the both option.

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

MPLS VPN architecture introduced the extended community BGP attribute. BGP still supports
the standard community attribute, which has not been superseded by extended communities.
The default community propagation behavior for standard BGP communities has not changed.
Community propagation still must be configured manually. Extended BGP communities are
propagated by default because their propagation is mandatory for successful MPLS VPN
operation.
The neighbor send-community command was extended to support standard and extended
communities. Use this command to configure propagation of standard and extended
communities if your BGP design relies on use of standard communities. An example would be
to propagate quality of service (QoS) information across the network.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-41

 Router (config-router)#
Ino bgp default ipv4-unicast

Cisco lOS and lOS XE Software only

The exchange of IPv4 routes between BGP neighbors is enabled by
default. Every configured neighborwill also receive IPv4 routes.
This command disables the default exchange of IPv4 routes. Neighbors
that need to receive IPv4 routes have to be activated for IPv4 route
exchange.
Use this command when the same router carries Internet and VPNv4
routes and you do not want to propagate Internet routes to some PE
neighbors.

C2012Ci:OC08rd'oritstrffili*'".A1lrigi'ts"'""'v.....

The BGP configuration that has been discussed so far is appropriate for situations where the PE
routers provide Internet and VPN connectivity. If the PE routers provide only VPN
connectivity, they do not need Internet routing, and the IPv4 route exchange should be disabled.
Here are the two ways of disabling IPv4 route exchange:
To disable IPv4 route exchange for only a few neighbors, your best option is to disable the
IPv4 route exchange on a neighbor-by-neighbor basis by using the no neighbor activate
command.
To disable IPv4 route exchange for most (or all) of the neighbors, you can use the no bgp
default ipv4-unicast command. After you enter this command, you must manually activate
IPv4 route exchange for each configured global BGP neighbor.

2-42 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Neighbor 172.16.32.14 receives only Internet routes.
Neighbor 172.16.32.15 receives only VPNv4 routes.
Neighbor 172.16.32.27 receives Internet and VPNv4 routes.

router bgp 65173

no bgp default ipv4-unicast
neighbor 172.16.32.14 remote-as 65173
neighbor 172.16.32.15 remote-as 65173
neighbor 172.16.32.27 remote-as 65173
! Activate IPv4 route exchange

address-family ipv4
neighbor 172.16.32.14 activate
neighbor 172.16.32.27 activate
! Step#2 - VPNv4 route exchange

address-family vpnv4
neighbor 172.16.32.15 activate
neighbor 172.16.32.27 activate

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts......".......

In this example, only a subset of BGP neighbors needs to receive IPv4 routes.
In the figure, the default propagation of IPv4 routes is therefore disabled. IPv4 route
exchange-and VPNv4 route exchange-is manually activated on a neighbor-by-neighbor
basis:
Neighbor 172.16.32.14 receives only Internet routes that are based on the IPv4 activation.
Neighbor 172.16.32.15 receives only VPNv4 routes that are based on the VPNv4
activation.
Neighbor 172.16.32.27 receives Internet and VPNv4 routes that are based on both IPv4 and
VPNv4 activations.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-43

 ------- MPLS VPN Backbone
CE-Y1

CE-X3 CE-Y3

interface loopback a interface loopback 0

ipv4 address 172.16.1.2 255.255.255.255 ip address 172.16.1.1255.255.255.255

router bgp 64500 router bgp 64500

address-family vpnv4 unicast neighbor 172.16.1.2 remote-as 64500
! neighbor 172.16.1.2 update-source loopback 0
neighbor 172.16.1.1 !
remote-as 64500 address-family vpnv4
update-source LoopbackO neighbor 172.16.1.2 activate
address-family vpnv4 unicast neighbor 172.16.1. 2 next-hop-self
next-hop-self neighbor 172.16.1.2 send-community both

C2012Ci'"""8rd'oritstrffili*'".A1lrigi'ts"'""'v.....

The right box in the figure shows Cisco lOS Software configuration. A neighbor must be
configured in the BGP section and then activated in the address family block. The extended
community command is added automatically.
The left box shows Cisco lOS XR Software configuration. In this case, the VPNv4 address
family is configured in the BGP section and then applied in the neighbor configuration block.

2-44 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary points that were discussed in this lesson.

5 g
A VRF table is a routing and forwarding instance that associates
additional attributes such as RD, import RT, and export RT to routing
entries.
"VRF-Lite" equals "VRF without the need to run MPLS."
MP-BGP is responsible for allocating labels for VPN routes and
advertising them to other edge routers when using MPLS.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-45

2-46 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 21

Connecting Customers Using

Simple Routing Protocols
Overview
This lesson explains provider edge (PE)-customer edge (CE) routing protocol configuration
steps and the various routing protocols that you can run between PE and CE routers. These
protocols include Routing Information Protocol (RIP), Enhanced Interior Gateway Routing
Protocol (EIGRP), and static routes.
It is important to understand not only what you can configure between PE and CE routers when
you are setting up Multiprotocol Label Switching (MPLS) VPNs, but also how to accomplish
the configuration successfully. It is also very important to be able to determine what steps you
should take when trying to solve a problem with your MPLS VPN network.

Objectives
Upon completing this lesson, you will be able to describe how to configure routing protocols
between PE and CE routers. You will be able to meet this objective:
Connect customers using per-VRF static routes, RIP PE-CE routing sessions, and EIGRP
PE-CE routing sessions
PE-CE Routing
This topic identifies the requirements for configuring PE-CE routing protocols.

E-
PE-CE routing protocols are configured for individual VRFs.
Cisco lOS and lOS XE Software
- Per-VRF routing protocols can be configured in two ways:
o Per-VRF parameters are specified in routing contexts, which are selected
with the address-family command.
o A separate OSPF process is started for each VRF.
Cisco lOS XR Software
- Per-VRF parameters are specified in the routing contexts.
- A separate OSPF process can also be configured for each VRF, but using
multiple instances of OSPF will use more router resources.

C2012Ci'""".r'lC!Iait. .ffili*'".A1lrigl'tsreoerv.....

After you configure virtual routing and forwarding (VRF) instances and establish Multiprotocol
Internal Border Gateway Protocol (MP-IBGP) connectivity between PE routers, you need to
configure routing protocols between the PE router and the attached CE routers. The PE-CE
routing protocols must be configured for individual VRFs. Sites that are in the same VPN but in
different VRFs cannot share the same PE-CE routing protocol.

Note The per-VRF configuration of the PE-CE routing protocols is another good reason for
grouping as many sites into a VRF as possible.

The per-VRF routing protocols can be configured in these ways:

Cisco lOS and lOS XE Software
Per-VRF routing protocols can be configured in two ways:
Per-VRF parameters are specified in routing contexts, which are selected with
the address-family command.
A separate OSPF process has to be started for each VRF.
Before Cisco lOS Release l2.3(4)T, the overall number of routing processes per
router was limited to 32, of which only 28 were available for VRF assignment.
Cisco lOS XR Software
Per-VRF parameters are specified in the routing contexts.
A separate OSPF process can also be configured for each VRF, but using multiple
instances of OSPF will use more router resources.

2-48 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config)#
router bgp as-number
Cisco lOS address-family ipv4 vrf vrf-name
and lOS XE ... Non-BGP redistribution

RPjOjRSPOjCPUO:Router(config)#
router bgp as-number
Cisco lOS XR vrf vrf-name
address-family ipv4 unicast
... Non-BGP redistribution

o Select the per-VRF BGP context with the address-family command.

o Configure CE External Border Gateway Protocol neighbors in VRF contexts,
not in global BGP configuration.
o All non-BGP per-VRF routes have to be redistributed into a per-VRF BGP
context to be propagated by MP-BGP to other PE routers.

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

On Cisco lOS and lOS XE devices, select the VRF routing context with the address-family
ipv4 vrf vrj-name command in the RIP and Border Gateway Protocol (BGP) routing processes.
All per-VRF routing protocol parameters (network numbers, passive interfaces, neighbors,
filters, and so on) are configured under this address family.
On Cisco lOS XR devices, first define a VRF with the vrf vrfname command in the BGP
routing processes. Then select the routing context with the address-family ipv4 unicast
command. All per-VRF routing protocol parameters (network numbers, passive interfaces,
neighbors, filters, and so on) are configured under this address family.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-49

 Router (config)#
ip route vrf vrf-name prefix mask [next-bop-address]
[interface interface-number]
This command configures per-VRF static routes.
The route is entered in the VRF table.
You must specify a next-hop IP address if you are not using a
point-to-point interface.

Sample router configuration:

ip route vrf Customer_ABC 10.0.0.0 255.0.0.0 10.250.0.2

!
router bgp 65173
address-family ipv4 vrf Customer ABC
redistribute static

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

To establish static routes for a VPN VRF instance on Cisco lOS and lOS XE devices, use the ip
route vrf command in global configuration mode. To disable static routes, use the no form of
this command.

~-1IIIIF

RP/O/RSPO/CPUO:Router(config)#
router static
vrf vrf-name
address-family ipv4 unicast
prefix mask [next-bop-address] [interface interface-number]

Sample router configuration:

router static
vrf Customer A
address-family ipv4 unicast
10.0.2.0124 192.168.0.1

router bgp 64500

vrf Customer A
rd 64500:1
address-family ipv4 unicast
redistribute static

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

On Cisco lOS XR devices, you must first define the static router. To enter static router
configuration mode, use the router static command in global configuration mode. Use
the vrf command to configure a VRF instance. Enter address family configuration mode with
the address-family ipv4 unicast command to configure a static route.

2-50 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ~-1IIIIF
MPLS VPN Backbone
AS 64500
CE-A1 CE-A2
Cisco lOS and Cisco lOS
lOS XE XR

ip route vrf Customer A 10.0.1.0 255.255.255.0

router bgp 64500 router static

address-family ipv4 vrf Customer A vrf Customer A
redistribute static address-family ipv4 unicast
no auto-summary 10.0.2.0/24 192.168.0.1

router bgp 64500

vrf Customer A
rd 64500:1
address-family ipv4 unicast
redistribute static

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The examples in the figure (Cisco lOS and lOS XE and lOS XR CLls) show how to configure
PE-CE routing using static routes.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-51

 A routing context is configured for each VRF running RIP.
RIP parameters have to be specified in the VRF.
Some parameters configured in the RIP process are propagated to
routing contexts (for example, RIP version).
Only RIPv2 is supported.

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

To configure RIP as the PE-CE routing protocol on Cisco lOS and lOS XE devices, start the
configuration of the individual routing context with the address-family ipv4 vrf vrj-name
command in router configuration mode. You can enter all standard RIP parameters in the per-
VRF routing context. Global RIP parameters that are entered in the scope of RIP router
configuration are inherited by each routing context and can be overwritten if needed in each
routing context.

Note Only RIP version 2 (RIPv2), and not RIP version 1 (RIPv1), is supported as the PE-CE
routing protocol. It is a good practice to configure the RIP version as a global RIP parameter
using the version 2 command in router configuration mode.

2-52 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Router (config)#
router rip
version 2
Cisco lOS
and lOS XE address-family ipv4 vrf vrf-name
redistribute bgp as-number metric transparent

Router (config)#
router rip
vrf vrf-name
Cisco lOS XR
redistribute bgp as-number
default-metric number-value

BGP routes must be redistributed back into RIP.

The RIP hop count must be manually set for routes that are redistributed into
RIP.
When you are using RIP with other protocols, you must set the metric manually.

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The interior gateway protocol (IGP) metric is always copied into the multi-exit discriminator
(MED) attribute of the BGP route when an IGP route is redistributed into BGP. Within
standard BGP implementation, the MED attribute is used only as a route selection criterion.
The MED attribute is not copied back into the IGP metric. The IGP metric must be specified in
the redistribute command or by using the default-metric command in router configuration
mode.
On Cisco lOS and lOS XE devices, the MPLS VPN extension to the redistribute command
(the metric transparent option) allows the MED attribute to be inserted as the IGP metric of a
route redistributed from BGP back into RIP. This extension gives transparent end-to-end (from
the customer perspective) RIP routing:
By default, the RIP hop count is inserted into the BGP MED attribute when the RIP route is
redistributed into BGP by the ingress PE router.
You can configure the value of the MED attribute (the original RIP hop count) to be copied
into the RIP hop count when the BGP route is redistributed back into RIP. This action
causes the whole MPLS VPN backbone to appear as a single hop to the CE routers.

Note You should not change the MED value within BGP if you use the redistribute metric
transparent command.

On Cisco lOS XR devices, use the default-metric command in VRF configuration mode for
RIP routing. The default metric value (number-value) is a number from 1 to 15.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-53

 MPLS VPN Backbone
CE-A1 AS 64500 CE-A2
Cisco lOS and Cisco lOS
lOS XE XR

router rip
router rip
vrf Customer A
version 2
interface GigabitEthernetOjOjOjO
address-family ipv4 vrf Customer A
redistribute bgp 64500 metric transparent
redistribute bgp 64500
network 10.0.0.0
defaul t-metric 5
no auto-summary

router bgp 64500

router bgp 64500
vrf Customer A
address-family ipv4 vrf Customer A
rd 64500:1
redistribute rip
address-family ipv4 unicast
no auto-summary
redistribute rip

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The examples in the figure (Cisco lOS, IOS XE, and lOS XR CLls) show how to configure PE-
CE routing using RIP as a routing protocol.

2-54 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Provides EIGRP with the capability to redistribute routes through a VPN
cloud.
EIGRP extended community attributes are used to define EIGRP routes
and preserve internal metrics.
Supports sao capabilities to filter VPN traffic.

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The MPLS VPN Support for EIGRP between Provider Edge and Customer Edge feature
provides the capability to transparently connect EIGRP customer networks through an MPLS-
enabled BGP core network. EIGRP routes can then be redistributed through the VPN across the
BGP network as internal BGP (IBGP) routes. The configuration of this feature does not require
any customer equipment upgrades or configuration changes. This feature is configured only on
PE routers within the service provider network.
Customer networks and remote sites are connected to each other through the MPLS VPN. The
configuration of this feature allows several EIGRP sites to connect seamlessly and appear as a
single network. This integration is transparent to the customer sites. When this feature is
enabled, EIGRP routes are converted to IBGP routes and transported through the BGP core
network. EIGRP extended community attributes are used to define EIGRP routes and preserve
internal metrics. These attributes are carried across the core network by multiprotocol BGP
(MP-BGP).
This feature also introduces EIGRP support for MPLS and BGP extended community attributes
such as Site of Origin (SaO).

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-55

 Router (config)#
router eigrp autonomous-system-numberl
Cisco lOS address-family ipv4 vrf vrf-name
and lOS XE autonomous-system as-number
redistribute bgp as-number metric metric-value

Router (config)#
router eigrp autonomous-system-numberl
vrf vrf-name
Cisco lOS address-family ipv4
XR autonomous-system as-number
redistribute bgp as-number metric metric-value

Enables the EIGRP AS number of the CE under the address family.

Configures per-instance AS number.
Configures router redistribution.
External routes that are received without the configured metric are not to be
advertised to the CE router.
C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The IGP metric is always copied into the MED attribute of the BGP route when an IGP route is
redistributed into BGP. Within a standard BGP implementation, the MED attribute is used only
as a route-selection criterion. The MED attribute is not copied back into the IGP metric. The
metric must be configured for routes from external EIGRP autonomous systems and non-
EIGRP networks before these routes can be redistributed into an EIGRP CE router. The metric
can be configured in the redistribute statement using the redistribute (IP) command or
configured with the default-metric (EIGRP) command.

Note In an MPLS VPN environment, the original EIGRP metrics must be carried inside MP-BGP
updates. This configuration is achieved by using BGP extended community attributes to
carry and preserve EIGRP metrics when crossing the MP-IBGP domain. These communities
define the intrinsic characteristics that are associated with EIGRP, such as the AS number
or EIGRP cost metric (bandwidth, delay, load, reliability, and MTU, for example).

2-56 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 MPLS VPN Backbone
AS 64500
CE-A1 CE-A2
Cisco lOS and Cisco lOS
lOS XE XR

router eigrp 1
router eigrp 1 vrf Customer A
address-family ipv4 vrf Customer A [ ... J address-family ipv4
[ J autonomous-system 1 default-metric 10000 100 255 1 1500
network 10.0.0.0 255.255.255.0 autonomous-system 1
redistribute bgp 64500 metric 10000 100 redistribute bgp 64500
255 1 1500 interface GigabitEthernetOjOjOjO
no auto-summary
router bgp 64500
router bgp 64500 vrf Customer A
address-family ipv4 vrf Customer A rd 64500:1
redistribute eigrp 1 metric 1 address-family ipv4 unicast
redistribute eigrp 1

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The examples in the figure (Cisco lOS, IOS XE, and lOS XR CLls) show how to configure PE-
CE routing using EIGRP as the routing protocol.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-57

 Site B
Site A P-Network ~JZJI'b EIGRP 101
EIGRP 101 AS 64500

CE-EIGRP-A 1

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

Routing loops and suboptimal routing generally occur because of mutual redistribution taking
place between EIGRP PE-CE and MP-BGP in an MPLS VPN environment. Routing loops can
occur in the following scenarios:
A route that is received by a multihomed site from the backbone through one link can be
forwarded back to the backbone through the other link.
A route that originated in a multihomed site and that was sent to the backbone through one
link can come back through the other link.
The figure shows an MPLS VPN network for a customer that has two sites, Site A and Site B.
Site B is multihomed. The figure shows that EIGRP route 10.1.2.0/24 received by the
multihomed site (Site B) is redistributed into the backbone at PE-Site-Y.
Routing loops and suboptimal routing can be avoided by using the following:
The BGP Cost Community feature, which can be used to force BGP to compare locally
originated EIGRP routes and MP-IBGP routes, based on the EIGRP metric
The EIGRP sao feature on PE and CE routers, which can be used to prevent routing loops
Note The sao attribute is needed only for customer networks with multihomed sites. Loops can
never occur in customer networks that have only stub sites.

2-58 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Site B
EIGRP 101

Cisco lOS and

CE-EIGRP-A1

;I llliiiiiiil
,PE.-S.ite-.Y

Cisco lOS and

10SXE 10SXR
route-map sao Support permit 10 router eigrp 1
vrf Customer 1
set extcommunity soo 64500:2
address-family ipv4
interface GigabitEthernetOjO autonomous-system 2
interface GigabitEthernetOjO
ip vrf forwarding Customer_A
ip vrf sitemap SaO_Support site-of-origin 64500:2

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The configuration of the SOO extended community allows routers that support this feature to
identify the site from which each route originated. When this feature is enabled, the EIGRP
routing process on the PE or CE router checks each received route for the SOO extended
community and filters based on the following conditions:
A received route from BGP or a CE router contains an SOO value that matches the SOO
value on the receiving interface:
If a route is received with an associated SOO value that matches the SOO value that
is configured on the receiving interface, the route is filtered out because it was
learned from another PE router or from a backdoor link. This behavior is designed to
prevent routing loops.
A received route from a CE router is configured with an SOO value that does not match:
If a route is received with an associated SOO value that does not match the SOO
value that is configured on the receiving interface, the route is accepted into the
EIGRP topology table so that it can be redistributed into BGP.
If the route is already installed in the EIGRP topology table but is associated with a
different SOO value, the SOO value from the topology table is used when the route
is redistributed into BGP.
A received route from a CE router does not contain an SOO value:
If a route is received without an SOO value, the route is accepted into the EIGRP
topology table. The SOO value from the interface that is used to reach the next-hop
CE router is appended to the route before it is redistributed into BGP.
When BGP and EIGRP peers that support the SOO extended community receive these routes,
they also receive the associated SOO values and pass them to other BGP and EIGRP peers that
support the SOO extended community. This filtering is designed to prevent transient routes
from being relearned from the originating site, which prevents transient routing loops from
occurnng.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-59

 In conjunction with BGP Cost Community, EIGRP, BGP, and the routing information base
(RIB) ensure that paths over the MPLS VPN core are preferred over backdoor links.
In the example in the figure, a route map, sao_support, with a specific sao value of 64500:2
is defined. The newly defined route map is then applied to VRF Customer-EIGRP-AI, which
connects the EIGRP neighbor (the CE-EIGRP-A2 router) to the PE-Site-Y router.

RPjOjRSPOjCPUO:PE3# show vrf all detail

VRF Customer_I; RD 1:210; VPN ID not set

Description not set
Interfaces:
Gigabi tEthernetO /0 / a/0
Address family IPV4 Unicast
Import VPN route-target communities:
RT,1,210
Export VPN route-target communities:
RT,1,210
No import route policy
No export route policy
<--- text omitted --->

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'tsreoerv.....

To display the set of defined VRFs, use the show ip vrf (lOS and lOS XE) or show vrf all
(lOS XR) command in EXEC mode.
To display the interfaces that are associated with a specific VRF, use the show ip vrf
interfaces (lOS and lOS XE) or show ipv4 vrf all interface brief (lOS XR) command.

2-60 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 o 172.16.1. 0/24 [110/2] via 192.168.101. 11, 1w6d, GigabitEthernetO /0 /0 /0
B 172.16.2.0/24 [200/2] via 10.2.1.1 (nexthop in vrf default) I lw6d
C 192.168.101.0/24 is directly connected, 2wOd, GigabitBthernetOjOjOjO
L 192.168.101.10/32 is directly connected, 2wOd, GigabitBthernetOjOjOjO
B 192.168.102.0/24 [200/0] via 10.2.1.1 (nexthop in vrf default) I lw6d

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

These two commands can be used to monitor VRF routing:

The show ip route vrf (IOS and lOS XE) or show route vrf (IOS XR) command displays
the VRF routing table.
The show ip bgp vpnv4 vrf (IOS and lOS XE) or show bgp vpnv4 unicast vrf (IOS XR)
command displays the VRF BGP table.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-61

 RPjOjRSPOjCPUO:PE1#show bgp neighbors
BGP neighbor is 10.0.1.1
Remote AS 64500, local AS 64500, internal link
Remote router ID 10.0.1.1
BGP state = Established, up for 2wOd
Last read 00:00:48, Last read before reset 00:00:00
Hold time is 180, keepalive interval is 60 seconds
Configured hold time: 180, keepalive: 60, min acceptable hold time: 3

<--- text omitted

Precedence: interne t
Neighbor capabilities:
Route refresh: advertised and received
4-byte AS: advertised and received
Address family VPNv4 Unicast: advertised and received

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The show ip bgp neighbors (Cisco lOS and lOS XE) or show bgp neighbors (Cisco lOS XR)
command is described in detail in the Cisco lOS and Cisco lOS XR Software documentation.
This command is used to monitor BGP sessions with other PE routers and the address families
that are negotiated with these neighbors.

2-62 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 RPjOjRSPOjCPUD:PE1#show bgp vpnv4 unicast rd 1:210
BGP router identifier 10.1.1.1, local AS number 64500
BGP generic scan interval 60 sees
<--- text omitted ---:>
Network Next Hop Metric LocPrf Weight Path
Route Distinguisher: 1:210 (default for vrf Customer_I)
'> 172 .16.1. 0/24 192.168.101.11 32768
'>il72 .16.2.0/24 10.2.1.1 100 0
'> 192.168.101.0/24 0.0.0.0 32768
'>il92 .168 .102.0/24 10.2.1.1 100 0

Processed 4 prefixes, 4 paths

C2012Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The show ip bgp vpnv4 (Cisco lOS and lOS XE) or show bgp vpnv4 unicast (Cisco lOS XR)
command displays IPv4 BGP information and VPNv4 BGP information. To display VPNv4
BGP information on devices that are running Cisco lOS or lOS XE, use one of these keywords:
all to display the whole contents of the VPNv4 BGP table
vrf vrfname to display VPNv4 information that is associated with the specified VRF
rd route-distinguisher to display VPNv4 information that is associated with the specified
RD

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-63

 =

RPjOjRSPOjCPUD:PE1#sh mpls forwarding vrf Customer_l

Tue Jan 3 12:19:19.574 UTe
Local Outgoing Prefix Outgoing Next Hop Bytes
Label Label or ID Interface Switched

16017 Unlabelled 172.16.1.0/24[V] GiO/O/O/O 192.168.101.11 500

16018 Aggregate Customer 1: Per-VRF Aggr[V]
Customer 1 500

C2012Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

These three commands can be used to display per-VRF forwarding information base (FIB) and
label forwarding information base (LFIB) structures:
The show ip cef vrf (lOS and lOS XE) or show cef vrf (lOS XR) command displays the
VRF FIB.
The show ip cef vrf detail (lOS and lOS XE) or show cef vrf detail (lOS XR) command
displays detailed information about a single entry in the VRF FIB.
The show mpls forwarding vrf (Cisco lOS, lOS XE, and lOS XR) command displays all
labels that are allocated to VPN routes in the specified VRF.

2-64 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Performs PE-CE Telnet through specified VRF
I telnet vrf vrf-name ip-address I
Performs ping based on VRF routing table
Iping vrf vrf-name ip-address I
Checks MPLS LSP connectivity
Iping mpls ipv4 destination-address

Performs VRF-based traceroute

Itrace vrf vrf-name ip-address

Discovers MPLS LSP routes

Itrace mpls ipv4 destination-address

These commands are the same in Cisco lOS, lOS XE, and lOS XR
Software.
C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

Additional Cisco lOS Software monitoring commands are as follows:

The telnet command can be used to connect to a CE router from a PE router using the Ivrf
option.
The ping vrf command can be used to ping a destination host reachable through a VRF.
The ping mpls command can be used to check MPLS label-switched path (LSP)
connectivity.
The trace vrf command can be used to trace a path toward a destination reachable through
aVRF.
The trace mpls command can be used to discover the MPLS LSP routes that packets
actually take when traveling to their destinations.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-65

Summary
This topic summarizes the primary point that was discussed in this lesson.

2 3
All routing protocols that support per-VRF routing can be used for route
exchange between the PE and CEo

C2012Ci'""""rd'oritstrffili*'".A1lrigl'ts......,'......

2-66 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 31

Connecting Customers Using

BGP or OSPF
Overview
This lesson explains the provider edge (PE)-customer edge (CE) advanced routing protocol
configuration steps and the various routing protocols that you can run between PE and CE
routers.
The Open Shortest Path First (OSPF) and Border Gateway Protocol (BGP) routing protocols
are discussed. This lesson describes the steps that are needed for Multiprotocol Label Switching
(MPLS) VPN troubleshooting.

Objectives
Upon completing this lesson, you will be able to describe how to configure and troubleshoot
routing protocols between PE and CE routers. You will be able to meet these objectives:
Configure an OSPF PE-CE routing session
Configure a BGP PE-CE routing session
Describe how to troubleshoot MPLS VPNs
OSPF as the PE-CE Routing Protocol
This topic describes OSPF as the routing protocol between PE and CE routers.

OSPF Area 0 (Backbone Area) OSPF divides a network into areas,

all of them linked through the
backbone (Area 0).
Areas could correspond to
Area Border Router
individual sites from an MPLS VPN
perspective.

Area 1

From the customer perspective, an

MPLS VPN-based network has a
BGP backbone with IGP running at BGP Backbone
customer sites.
Redistribution between IGP and
BGP is performed to propagate
PE Router
customer routes across the MPLS
VPN backbone.

CE Router Site IGP Site IGP Site IGP

C2011Ci'""".r'lC!Iait. .ffili*'".A1lrigl'tsreoerv.....

The OSPF routing protocol was designed to support hierarchical networks with a central
backbone. A network that is running OSPF is divided into areas. All areas must be directly
connected to the backbone area (Area 0). The whole OSPF network (backbone area and other
connected areas) is called the OSPF domain.
The OSPF areas in the customer network can correspond to individual sites, but these other
options are often encountered:
A single area could span multiple sites (for example, the customer decides to use one area
per region, but the region contains multiple sites).
The backbone area could be extended into individual sites.
The MPLS VPN routing model introduces a BGP backbone into the customer network. Isolated
copies of the interior gateway protocol (IGP) run at every site, and Multiprotocol BGP
(MP-BGP) is used to propagate routes between sites. Redistribution between the customer IGP
(running between PE routers and CE routers) and the backbone MP-BGP is performed at every
PE router.

2-68 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 . - - - - - - - - l 2.An OSPF route is redistributed into BGP.

BGP Backbone

4. The MP-BGP route is

redistributed into OSPF.

---,.,....---l 5. The OSPF route is

propagated as an
external route into
other sites.

Area 1 Area 2 Area 3

' - - - - - - - - - l 1. A local subnetwork is announced to the PE router as

type 1 or type 2 LSA.

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The IGP-BGP redistribution that is introduced by the MPLS VPN routing model does not fit
well into customer networks running OSPF. When an OSPF customer is migrated to an MPLS
VPN implementation, any route that is redistributed into OSPF from another routing protocol
will now be redistributed as an external OSPF route. The OSPF routes received by one PE
router are propagated across the MPLS backbone and redistributed back into OSPF at another
site as external OSPF routes.

Note Remember that link-state advertisement (LSA) 1 and LSA 2 never leave the OSPF area.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-69

 The OSPF route type is not preserved when the OSPF route is
redistributed into BGP.
All OSPF routes from a site are inserted as external (type 5 LSA) routes
into other sites.
The result is that OSPF route summarization and stub areas are hard to
implement.
Conclusion: MPLS VPNs must extend the classic OSPF-BGP routing
model.

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

With the traditional OSPF-BGP redistribution, the OSPF route type (internal or external route)
is not preserved when the OSPF route is redistributed into BGP. When that route is
redistributed back into OSPF, it is always redistributed as an external OSPF route.
This list identifies some of the caveats that are associated with external OSPF routes:
External routes cannot be summarized.
External routes are flooded across all OSPF areas.
External routes could use a different metric type that is not comparable to OSPF cost.
External routes are not inserted in stub areas or not-sa-stubby areas (NSSAs).
Internal routes are always preferred over external routes, regardless of their cost.

Because of all these caveats, migrating an OSPF customer toward an MPLS VPN service might
have a severe impact on the routing of that customer. The MPLS VPN architecture must
therefore extend the classic OSPF-BGP routing model to support transparent customer
migration.

2-70 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 OSPF Area 0 might extend into individual sites.
The MPLS VPN backbone has to become a superbackbone for OSPF.
BGP Backbone

OSPF between sites will not use normal OSPF-BGP redistribution.

OSPF continuity must be provided across the MPLS VPN backbone:
- Internal OSPF routes should remain internal OSPF routes.
- External routes should remain external routes.
- OSPF metrics should be preserved.
CE routers run standard OSPF software.
C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The MPLS VPN architecture extends the OSPF architecture by introducing another backbone
above OSPF Area 0, the superbackbone. The OSPF superbackbone is implemented with
MP-BGP between the PE routers but is otherwise transparent to the OSPF routers. The
architecture even allows disjointed OSPF backbone areas (Area 0) at MPLS VPN customer
sites.
These goals must be met by the OSPF superbackbone:
The superbackbone will not use standard OSPF-BGP redistribution.
OSPF continuity must be provided between OSPF sites:
Internal OSPF routes must remain internal OSPF routes.
External OSPF routes must remain external OSPF routes.
Non-OSPF routes that are redistributed into OSPF must appear as external OSPF
routes in OSPF.
OSPF metrics and metric types (externall or external 2) must be preserved.
The OSPF superbackbone will be transparent to the CE routers that run standard OSPF
software.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-71

 .-------------l 2. The PE router propagates the route into the
superbackbone. Route summarization can be
performed on the area boundary.

3. The route from the superbackbone

is inserted into other areas as an
interarea route.

Area 0

' - - - - - - - - l 1. A local subnetwork is announced to the PE

router as type 1 or type 2 LSA.

Extended BGP communities are used to propagate OSPF route types across the
BGP backbone.
OSPF cost is copied into the MED attribute.

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The MPLS VPN superbackbone appears as another layer of hierarchy in the OSPF architecture.
The PE routers that connect regular OSPF areas to the superbackbone therefore appear as OSPF
Area Border Routers (ABRs) in the OSPF areas to which they are attached. ABRs also appear
as Autonomous System Boundary Routers (ASBRs) in nonstub areas.
From the perspective of a standard OSPF-speaking CE router, the PE routers insert interarea
routes from other areas into the area in which the CE router is present. The CE routers are not
aware of the superbackbone or of other OSPF areas present beyond the MPLS VPN
superbackbone.
With the OSPF superbackbone architecture, the continuity of OSPF routing is preserved:
The OSPF intra-area route, described in the OSPF router LSA or network LSA, is inserted
into the OSPF superbackbone by redistributing the OSPF route into MP-BGP. Route
summarization can be performed on the redistribution boundary by the PE router.
The MP-BGP route is propagated to other PE routers and inserted as an OSPF route into
other OSPF areas.

The OSPF superbackbone is implemented with the help of several BGP attributes:
A new BGP extended community was defined to carry the OSPF route type and OSPF area
across the BGP backbone.
As in the standard OSPF-BGP redistribution, the OSPF cost is carried in the multi-exit
discriminator (MED) attribute.

2-72 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Internal OSPF routes
The OSPF route type is copied into the
extended BGP community on redistribution
into BGP.
The egress PE router performs interarea
transformation.

External OSPF routes

Routes are propagated in the same
way as internal OSPF routes across
the superbackbone.
The external metric and route type
are preserved. BGP
Backbone
Routes from other routing protocols
Routes from the MP-BGP backbone that did not
originate in OSPF are still subjectto standard
redistribution behavior when inserted into OSPF.

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The first diagram in the figure illustrates the propagation of internal OSPF routes across the
MPLS VPN superbackbone.
In the second diagram, the external OSPF routes are redistributed into MP-BGP in the same
way as the internal OSPF routes.
In the third diagram, the MPLS VPN superbackbone retains the traditional OSPF-BGP route
redistribution behavior for routes that did not originate in OSPF at other sites.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-73

 Follow these steps to configure OSPF as the PE-CE routing
protocol:
Configure a per-VRF copy of OSPF.
Configure redistribution of MP-BGP into OSPF.
Configure redistribution of OSPF into MP-BGP.

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

To configure OSPF as a PE-CE routing protocol, you must start a separate OSPF process for
each virtual routing and forwarding (VRF) instance in which you want to run OSPF. The per-
VRF OSPF process is configured in the same way as a standard OSPF process. You can use all
the OSPF features that are available in Cisco lOS Software.
You need to redistribute OSPF routes into BGP and redistribute BGP routes into OSPF if
necessary. Alternatively, you can originate a default route into a per-VRF OSPF process by
using the default-information originate always command in router configuration mode.
MP-BGP propagates more than just OSPF cost across the MPLS VPN backbone. The
propagation of additional OSPF attributes into MP-BGP is automatic and requires no extra
configuration.

2-74 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 router (config) #

router ospf process-id vrf vrf-name

... Standard OSPF parameters ...
This command starts the per-VRF OSPF routing process.
router (config-router) #
Iredistribute bgp as-number subnets
This command redistributes MP-BGP routes into OSPF. The
subnets keyword is mandatory for proper operation.
router (config) #
router bgp as-number
address-family ipv4 vrf vrf-name
redistribute ospf process-id [match [internal]
[external-l] [external-2]]
OSPF-BGP route redistribution is configured with the redistribute
command under the proper address-family command.
C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

In an MPLS VPN deployment, each VPN VRF needs a separate OSPF process when
configured to run OSPF. OSPF support for unlimited software VRFs per PE router addresses
the scalability issue for OSPF VPNs by eliminating the OSPF VPN limit of 32 processes.
To configure an OSPF routing process within a VRF on Cisco lOS and lOS XE devices, use
the router ospf command in global configuration mode. OSPF-BGP route redistribution is
configured with the redistribute command under the proper address-family command.
Without the OSPF match keyword specified, only internal OSPF routes are redistributed into
BGP.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-75

 ~-

RPjOjRPOjCPUO:router(config-ospf)#
I vrf vrf-name
... Standard OSPF parameters ...

This command starts the per-VRF OSPF routing process.

RPjOjRPOjCPUO:router(config-ospf-vrf)#
I redistribute bgp as-number
This command redistributes MP-BGP routes into OSPF.

RPjOjRPOjCPUO:router(config)#
router bgp as-number
vrf vrf-name
address-family ipv4 unicast
redistribute ospf process-id [match {external [112] 1
internal}]

OSPF-BGP route redistribution is configured under the proper

address-family command.
C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

To configure an OSPF routing process within a VRF on Cisco lOS XR devices, you must first
enter VRF configuration mode for OSPF routing. OSPF-BGP route redistribution is configured
with the redistribute command under the proper address-family command.

.----1 2. The OSPF route is received by a PE router,

redistributed into MP-BGP, and propagated
across the MPLS VPN backbone.
BGP Backbone

PE Router

... PE Router
5. The other PE router
would redistribute the
route back into BGP.

L..---::;;::;::-14. The OSPF route

is propagated
across the area.

Area 1 Area 2
1. The local subnetwork is announced to the PE router.

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

OSPF developers took many precautions to avoid routing loops between OSPF areas. For
example, intra-area routes are always preferred over interarea routes. These rules do not work
when the superbackbone is introduced. For example, consider the network in the figure, where
the receiving OSPF area has two PE routers attached to it.

2-76 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 A down bit has been introduced in the options field of the OSPF LSA header.
PE routers set the down bit when redistributing routes from MP-BGP into OSPF.
PE routers never redistribute OSPF routes with the down bit set into MP-BGP.
2. An OSPF route is received by a PE router, redistributed into
MP-BGP, and propagated across the MPLS VPN backbone.
BGP Backbone
.--- ~ 3. The route from the superbackbone is inserted
as the interarea route.

l
The route is never redistributed
back into the MP-BGP backbone.

L_.~_J14. The OSPF route is propagated

with the down bit set.

Area 1 Area 2

1. The local subnetwork is announced without the down bit.

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The down bit was introduced as a mechanism to prevent route redistribution loops between
OSPF and MP-BGP that is running between PE routers. The OSPF down bit is part of the
Options field in the OSPF LSA header.
The down bit is used between the PE routers to indicate which routes were inserted into the
OSPF topology database from the MPLS VPN superbackbone. Those routes thus will not be
redistributed back into the MPLS VPN superbackbone. The PE router that redistributes the
MP-BGP route as an OSPF route into the OSPF topology database sets the down bit. The other
PE routers use the down bit to prevent this route from being redistributed back into MP-BGP.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-77

 2. The OSPF route is propagated with the down 3. Because of administrative distances,
bit set. an OSPF route is preferred over an
MP-IBGP route. Packet flow across
the network is not optimal.
BGP Backbone

Another OSPF or
Area 1 Area 2 Non-OSPF Site
1. The OSPF route is received by a PE router
and redistributed into MP-BGP and OSPF.

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The OSPF superbackbone implementation with MP-BGP has other implications beyond the
potential for routing loops between OSPF and BGP. For example, consider the network in the
figure, which shows a typical flow for routing updates.

2-78 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 1. The OSPF route is propagated with the down 2. The OSPF route is ignored because
bit set. the down bit is set.

BGP Backbone

Another OSPF or
Area 1 Area 2 Non-OSPF Site
I Packet flow across the network is optimal.
C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

To prevent customer sites from acting as transit parts of the MPLS VPN network, the OSPF
route-selection rules in PE routers need to be changed. The PE routers must ignore all OSPF
routes with the down bit set, because these routes originated in the MP-BGP backbone, and the
MP-BGP route should be used as the optimum route toward the destination.
This rule is implemented with the routing bit in the OSPF LSA. For routes with the down bit
set, the routing bit is cleared and these routes never enter the IP routing table, even if they are
selected as the best routes by the Shortest Path First (SPF) algorithm.

Note The routing bit is the Cisco extension to OSPF and is used only internally in the router. The
routing bit is never propagated between routers in LSA updates.

With the new OSPF route-selection rules in place, packet forwarding in the network that is
shown in the figure follows the desired path.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-79

 =-...
OSPF prefers intra-area paths to interarea paths.
The path over a backdoor link will always be selected.
A sham link is a logical intra-area link.
It is carried by the superbackbone.
A sham link is required only
between two VPN sites High-Bandwidth
BGP Backbone
that belong to the same
area and have a backdoor
link for backup purposes.
OSPF adjacency is
established across the
sham link.
LSA flooding occurs
across the sham link.

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

Although OSPF PE-CE connections assume that the only path between two client sites is across
the MPLS VPN backbone, backdoor paths between VPN sites may exist.
The figure shows backdoor paths between VPN sites. If these sites belong to the same OSPF
area, the path over a backdoor link will always be selected, because OSPF prefers intra-area
paths to interarea paths. (PE routers advertise OSPF routes learned over the VPN backbone as
interarea paths.) For this reason, OSPF backdoor links between VPN sites must be taken into
account so that routing is performed based on policy.
Because each site runs OSPF within the same Area 1 configuration, all routing between the
sites follows the intra-area path across the backdoor links rather than over the MPLS VPN
backbone.
A sham link is required between any two VPN sites that belong to the same OSPF area and
share an OSPF backdoor link. If no backdoor link exists between the sites, no sham link is
required.

2-80 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 :1
5
_ _

2. The site 1 PE redistributes 3. The site 2 PE receives the OSPF type 1

the OSPF route into MP- LSA for the selected route from two
BGP because the selected directions. The OSPF cost of the sham
OSPF route was not High-Bandwidth link has been configured so that the
received via a sham link. sham link is preferred.
BGP Backbone

4. The site 2 PE is not

redistributing the
selected OSPF
route into MP-BGP
because the
preferred route was
received via a sham
link.

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

Because the sham link is seen as an intra-area link between PE routers, an OSPF adjacency is
created, and a database exchange (for the particular OSPF process) occurs across the link.
After the desired intra-area connectivity is created, the PE routers prefer routes that are learned
via the high-bandwidth backbone, because the OSPF cost of the sham link has been configured
so that it is preferred over the backdoor link. The implementation results in optimal packet
flow.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-81

 S L-.
A separate /32 address space is BGP Backbone AS 64500
required in each PE router for each
sham link. PE Router

This /32 address space:

- Is required so that OSPF packets can
be sent over the VPN backbone to the
remote end of the sham link
- Must belong to the VRF
- Must not be advertised by OSPF
- Must be advertised by BGP

Router (config-router) # Cisco lOS and lOS XE

area area-id sham-link source-address destination-address cost number

RPjOjRPOjCPUO: router (config-ospf)# Cisco lOS XR

vrf vrf-name
area area-id
sham-link source-address destination-address cost number

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

When you are configuring a sham link, a separate /32 address space is required in each PE
router.
These criteria apply to this /32 address space:
Required so that OSPF packets can be sent over the VPN backbone to the remote end of the
sham link
Must belong to the VRF
Must not be advertised by OSPF
Must be advertised by BGP

To configure a sham-link interface on a PE router that is running Cisco lOS and lOS XE
Software in an MPLS VPN backbone, use the area sham-link cost command in global
configuration mode.
To configure a sham-link interface using Cisco lOS XR Software, you must first enter VRF
configuration mode for OSPF routing and configure an area for the OSPF process using the
area command. Then you must assign interfaces to the sham link using the sham-link
command.

2-82 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
BGP as the PE-CE Routing Protocol
This topic describes BGP as the routing protocol between PE and CE routers.

Router (config) #
router bgp as-number
Cisco lOS address-family ipv4 vrf vrf-name
and lOS XE ... Per-VRF BGP definitions

RPjOjRPOjCPUO:Router(config)#

router bgp as-number

Cisco lOS vrf vrf-name
XR address-family ipv4 unicast
... Per-VRF BGP definitions

o Select a per-VRF BGP context with the address-family command.

o Configure CE EBGP neighbors in the VRF context, not in the global
BGP configuration.
o CE neighbors must be activated with the neighbor activate command.

When you configure BGP as the PE-CE routing protocol, you must start with the per-VRF BGP
configuration. Use the address-family ipv4 vrf vrf-name command in router configuration
mode on Cisco lOS and lOS XE devices. Enter address-family configuration mode, and then
define and activate the BGP neighbors. You also need to configure redistribution from all other
per-VRF routing protocols into BGP.
On Cisco lOS XR devices, first define the VRF with the vrf vr:f-name command in the BGP
routing processes. Then select the routing context with the address-family ipv4 unicast
command. All per-VRF routing protocol parameters (network numbers, passive interfaces,
neighbors, filters, and so on) are configured under this address family.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-83

 ~-
-~
...
router bgp 64501
neighbor 10.0.1.1 remote-as 64500
network 10.1.1.0 mask 255.255.255.0
MPLS VPN Backbone CE-BGP-A2
'-----,------------' AS 64500

CE-BGP-A1

router bgp 64500

ip vrf Customer_A address-family vpnv4 unicast
rd 64501:1 vrf Customer A
route-target both 64500:1 rd 64500:1
address-family ipv4 unicast
router bgp 64500
address-family ipv4 vrf Customer_A neighbor 10.2.1.1
neighbor 10.1.1.1 remote-as 64501 remote-as 64502
neighbor 10.1.1.1 activate update-source GigabitEthernetOjOjOjO
address-family ipv4 unicast

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The figure shows that BGP is activated on the CE-BGP-AI router and that the PE-X router is
defined as a BGP neighbor. In addition, on the PE-X router, the CE-BGP-Al router is defined
as a BGP neighbor and is activated under the address-family ipv4 vrf Customer_A command.
Both routers are running Cisco lOS and lOS XE Software.
You can also see that the PE-Y router is running Cisco lOS XR Software. Because all CE
routers in the example are running Cisco lOS and lOS XE Software, the configuration for the
CE-BGP-A2 and CE-BGP-A3 routers is omitted.

2-84 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Service providers offering MPLS VPN services are at risk of denial-of-
service attacks similar to those aimed at service providers offering BGP
connectivity:
- Any customer can generate any number of routes, using resources in the PE
routers.
- Therefore, the resources that are used by a single customer have to be
limited .
Cisco lOS Software offers two solutions:
- You can limit the number of routes received from a BGP neighbor.
- You can limit the total number of routes in a VRF.

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

MPLS VPN architecture achieves a tight coupling between the customer and the service
provider network, resulting in a number of advantages. The tight coupling could also result in a
few disadvantages, because the service provider network is exposed to design and configuration
errors in customer networks, and a number of new denial-of-service (DoS) attacks are based on
routing protocol behavior.
To limit the effect of configuration errors and malicious users, Cisco lOS Software offers two
features that limit the number of routes and the resource consumption that are available to a
VPN user at a PE router:
The BGP Maximum-Prefix feature limits the number of routes that an individual BGP peer
can send.
The VRF route limit restricts the total number of routes in a VRF regardless of whether
those routes are received from CE routers or from other PE routers via Multiprotocol
Internal Border Gateway Protocol (MP-IBGP).

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-85

 Router (config-router-af) #
Cisco lOS . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
and lOS neighbor ip-address maximum-prefix maximum [threshold]
XE [warning-only]

RP/O/RPO/CPUO:Router(config-bgp-nbr-af)#
Cisco
IOSXR maximum-prefix maximum [threshold] [warning-only]

Control how many prefixes can be received from a neighbor.

Optional threshold parameter specifies the percentage where a warning
message is logged (the default is 75 percent).
Optional warning-only keyword specifies the action on exceeding the
maximum number (the default is to drop peering).

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

To control how many prefixes can be received from a neighbor, use the maximum-prefix
command for the peer for the appropriate address family.
Cisco lOS and lOS XE Software:
neighbor {ip-address I peer-group-name} maximum-prefix maximum [thresholdJ
[restart restart-interval] [warning-only]
Cisco lOS and lOS XE Software:
maximum-prefix maximum [thresholdJ [warning-only]

2-86 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 The VRF maximum routes limit command limits the number of routes that are
imported into a VRF:
- Routes coming from CE routers
- Routes coming from other PE routers (imported routes)
The route limit is configured for each VRF.
If the number of routes exceeds the route limit:
- A syslog message (Cisco lOS and lOS XE Software) is generated.
- A SNMP trap (Cisco lOS XR Software) is generated.
- Cisco lOS, lOS XE, and lOS XR Software can be configured to reject routes (optional).

Router (config-vrf)#
Cisco lOS
and lOS XE maximum routes limit {warn-threshold warn-only}

RPjOjRSPOjCPUO:Router(config-vrf-af)#
Cisco lOS
XR Imaximum prefix limit [threshold]

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

The VRF route limit, unlike the BGP maximum-prefix limit, limits the overall number of routes
in a VRF regardless of their origin. As with the BGP Maximum-Prefix feature, the network
operator might be warned by a syslog message or Simple Network Management Protocol
(SNMP) trap when the number of routes exceeds a certain threshold. Additionally, you can
configure Cisco lOS and lOS XE and Cisco lOS XR Software to ignore new VRF routes when
the total number of routes exceeds the maximum configured limit.
The route limit is configured for each individual VRF, providing maximum design and
configuration flexibility.

Note The per-VRF limit could be used to implement add-on MPLS VPN services. A user wanting
a higher level of service might be willing to pay to be able to insert more VPN routes into the
network.

To limit the maximum number of routes in a VRF instance to prevent a PE router from
importing too many routes, use the maximum routes command in VRF configuration mode
(lOS and lOS XE Software) or VRF address family configuration mode (lOS XR Software).

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-87

 Customer A
AS 64501

VPN-IPv4 Update:
RD:192.168.61.0/24
RT= 64500:2

CE-BG~~'~':,';:'O"J ; - \
PE-Y

Cisco lOS and lOS XE ~ -\ Cisco lOS XR

,..------'-----------,
ip vrf Customer_A vrf Customer A
rd 64500,2 address-family ipv4 unicast
route-target both 64500:2 import route-target 64500:2
maximum routes 4 75 export route-target 64500:2
maximum prefix 4 75

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

The network designer can decide to limit the number of routes in a VRF.

In the figure, the network designer has decided to limit the number of routes in a VRF to four,
with the warning threshold being set at 75 percent (or three routes).
When the first two routes are received and inserted into the VRF, the router accepts them.
When the third route is received, a warning message is generated, and the message is repeated
with the insertion of the fourth route.
When the PE router receives the fifth route, the maximum route limit is exceeded, and the route
is ignored. The network operator is notified through another syslog message.

2-88 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ~ "7~

The customer wants to reuse an AS number on several sites:

CE-BGP-A1 announces network 10.1.0.0/16 to PE-Site-X.
The prefix announced by CE-BGP-A1 is propagated to PE-Site-Y as an
internal route through MP-BGP.
PE-Site-Y prepends AS 64500 to the AS path and propagates the prefix to
CE-BGP-A2.
CE-BGP-A2 drops the update because AS 64501 is already in the AS path.

Site A P-Network SiteS

AS 64501 AS 64500 AS 64501

WI II. glMili@ ,~ _I II. glMili@ ' . I II

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

Here are the two ways that an MPLS VPN customer can deploy BGP as the routing protocol
between PE and CE routers:
If the customer has previously used any other routing protocol in the traditional overlay
VPN network, there are no limitations on the numbering of the customer autonomous
systems. Every site can be a separate autonomous system (AS).
If the customer has used BGP as the routing protocol before, there is a good chance that all
the sites (or a subset of the sites) are using the same AS number.
BGP loop-prevention rules disallow discontiguous autonomous systems. Two customer sites
with the identical AS number cannot be linked by another AS. If such a setup happens (as in the
example in the figure), the routing updates from one site are dropped when the other site
receives them. There is no connectivity between the sites.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-89

 New AS path update procedures have been implemented to reuse an AS
number on all VPN sites.
The procedures allow the use of private and public AS numbers.
The same AS number may be used for all sites.
With as-override configured, the AS path update procedure on the PE router is
as follows:
- If the first AS number in the AS path is equal to the neighboring AS, it is replaced with
the provider AS number.
- If the first AS number has multiple occurrences (because of AS path prepend), all
occurrences are replaced with the provider AS number.
- After this operation, the provider AS number is prepended to the AS path.

Router (config-router-af) #
Cisco lOS
and lOS XE IneighbOr ip-address as-override
RPjOjRPOjCPUO:router (config-bgp-vrf-nbr-af) #
Cisco lOS
XR I as-override
C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

When you are migrating customers from traditional overlay VPNs to MPLS VPNs, it is not
uncommon to encounter a customer topology that requires a customer AS number to be used at
more than one site. This requirement can cause problems with the loop-prevention rules of
BGP. However, the AS-path update procedure in BGP has been modified to address this issue.
The new AS-path update procedure supports the use of one AS number at many sites (even
between several overlapping VPNs) and does not rely on a distinction between private and
public AS numbers.
The modified AS-path update procedure is called AS override:
The procedure is used only if the first AS number in the AS path is equal to the AS number
of the receiving BGP router.
In this case, all leading occurrences of the AS number of the receiving BGP router are
replaced with the AS number of the sending BGP router. Occurrences that are farther down
the AS path of the AS number of the receiving router are not replaced because they indicate
a real routing information loop.
An extra copy of the sending router AS number is prepended to the AS path. The standard
AS number prepending procedure occurs on every External Border Gateway Protocol
(EBGP) update.
To configure a PE router to override a site AS number with a provider AS number, use the
as-override command in the appropriate address family.

2-90 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ---_.~
PE-Site-YreplacesAS 64501 with AS 64500 in theAS path, prepends
another copy of AS 64500 to the AS path, and propagates the prefix.

Cisco lOS and lOS XE Cisco lOS XR

router bgp 64500
router bgp 64500
vrf Customer 2
address-family ipv4 vrf Customer_A
neighbor 10.1.1.1
neighbor 10.1.1.1 remote-as 64501
remote-as 64501
neighbor 10.1.1.1 activate
address-family ipv4 unicast
neighbor 10.1.1.1 activate
as-override

PE-X PE-Y

Mull Igi-ilM ' . Muli.l.liiMiI!Oi'] 'i!!"i'l

I.. t
C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

In this figure, customer sites A and B use BGP to communicate with the MPLS VPN backbone.
Both sites use AS 64501. Site B would drop the update that was sent by site A without the
AS-override mechanism.
The AS-override mechanism, configured on the PE-Site-Y router, replaces the customer AS
number (64501) with the provider AS number (64500) before sending the update to the
customer site. An extra copy of the provider AS number is prepended to the AS path during the
standard EBGP update process.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-91

 The BGP route is rejected because the PE3 router sees its own AS number
in the AS path.
Customer A:
Customer A: VPN
Site Spoke 1 EBGP Update VPN Hub Site
as-path (64501) AS 64503
AS 64501 CE3

jVRF' >

-------/7
CE2 EBGP Update
as-path
(1 64503 1 64501
router BGP 1
address-family IPv4 VRF Customer 1
neighbor CE4 allowas-in
C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

Consider a hub-and-spoke scenario that requires you to permit the routes that are coming from
the VRF hub site to re-enter the AS of the service provider. To do so requires that the spoke-to-
spoke communication happen through the VRF hub site.
The hub site connects to the provider with two links, which belong to two different VRFs on
PE3. One link is used to send updates to the hub site, and one link is used to receive updates
from the hub site. For BGP, this setup implies that a route traverses the service provider AS
from a VRF spoke site to the VRF hub site and traverses it again on the way to another VRF
spoke site. The PE3 router that connects to the VRF hub site sees its own AS number in the AS
path, so the BGP route is rejected.
To disable the AS-path loop check, you can configure the command neighbor allowas-in
number on the PE3 router that connects to the VRF hub site. The allowas-in command permits
multiple occurrences of the same AS number (in this case, the AS number of the service
provider) as the AS number of the BGP speaker in the AS path without BGP denying the route.
You can configure a number from 1 to 10 to specify the number of times that the AS number is
allowed in the AS path.

2-92 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 AS path-based BGP loop prevention is bypassed with the as-override and
allowas-in features. ."Ii'i@MiIllOi'] 'iIl!Oi,]-
Site B AS 64501

CE-BGP-A2

'Ii i U-@''+
ii'Ii,i@MiIl!Oi"-+
Sets the SOO value for a BGP neighbor
- , Ii i U-@ ["iIl@I,.

Cisco lOS Router (config-router-af) #

and lOS XE neighbor ip-address soo I AS:nn

Cisco lOS RPjOjRPOjCPUO:Router(config-bgp-vrf-nbr-af)#

XR Isite-oi-origin AS:nn

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

Most aspects ofBGP loop prevention are bypassed when you use either the as-override or
allowas-in command. Routing information loops can still be detected by manually counting
occurrences of an AS number in the AS path in an end-to-end BGP routing scenario and then
ensuring that the number field in the allowas-in command is set low enough to prevent loops.
The ability to continue to detect loops can present a particular problem when BGP is mixed
with other PE-CE routing protocols. The Site of Origin (SaO) extended BGP community can
be used as an additional loop-prevention mechanism in these situations.
The sao uniquely identifies the site that originates a route. It is a BGP extended community
that prevents routing loops or suboptimal routing, specifically when a back door is present
between VPN sites. The sao provides loop prevention in networks with dual-homed or
multihomed sites (sites that are connected to two or more PE routers). You can use it when an
IGP is the PE-CE routing protocol. You can also use it when BGP is used between PE and CE,
when the AS-path loop prevention cannot be trusted anymore. This situation happens when
BGP uses as-override or allowas-in. If the sao is configured for a CE router and a VPNv4
route is learned with the same sao, the route must not be put in the VRF routing table on the
PE and advertised to the CEo
Use this command to set the sao value for a BGP neighbor. The sao value is set under
address-family IPv4 VRF configuration mode either directly for a neighbor or for a BGP peer
group.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-93

Troubleshooting MPLS VPNs
This topic describes steps that are needed for MPLS VPN troubleshooting.

Perform basic MPLS troubleshooting:

Is Cisco Express Forwarding enabled?
Are labels for IGP routes generated and propagated?
Are large labeled packets propagated across the MPLS backbone
(maximum transmission unit issues)?

C2011Ci'""".r'lC!Iait. .ffili*'".A1lrigl'tsreoerv.....

Before you start in-depth MPLS VPN troubleshooting, you should ask the following standard
MPLS troubleshooting questions:
Is Cisco Express Forwarding enabled on all routers in the transit path between the PE
routers?
Are labels for BGP next hops generated and propagated?
Are there any maximum transmission unit (MTU) issues in the transit path (for example,
LAN switches not supporting a jumbo Ethernet frame)?

MPLS VPN troubleshooting consists of these two major steps:

Verifying the routing information flow, using the checks that are outlined in the figure
Verifying the data flow, or packet forwarding

2-94 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 6. Are VPNv4 routes redistributed
from BGP into the PE-CE routing
protocol?
CE-Spoke

4. Is the BGP route selection

process working correctly?

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

Verification of the routing information flow should be done systematically, starting at the
ingress CE router and moving to the egress CE router.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-95

 show bgp vpnv4 vrf vrf-name ip-prefix
debug bgp

show route vrf vrf-name

show bgp vpnv4 unicast vrf vrf-name ip-prefix

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

Troubleshooting routing information flow requires the verification of end-to-end routing

information propagation between CE routers.
Step 1 The first step is to check the routing information exchange from the CE routers to
the PE routers. Use the show ip route vrf vrJ-name (Cisco lOS and lOS XE) or
show route vrf vrj-name (Cisco lOS XR) command to verify that the PE router
receives customer routes from the CE router. Use traditional routing protocol
troubleshooting if needed.
Step 2 The CE routes that are received by the PE router need to be redistributed into MP-
BGP; otherwise, the routes will not be propagated to other PE routers. Proper
redistribution of CE routes into a per-VRF instance of BGP can be verified with the
show ip bgp vpnv4 vrf vrJ-name (IOS and lOS XE) or show bgp vpnv4 vrf vrj-
name (IOS XR) command. The route distinguisher (RD) prepended to the IPv4
prefix and the route targets (RTs) attached to the CE route can be verified with the
show ip bgp vpnv4 vrf vrJ-name ip-prejix (IOS and lOS XE) or show bgp vpnv4
vrf vrJ-name ip-prejix (IOS XR) command.
Step 3 The CE routes that are redistributed into MP-BGP need to be propagated to other PE
routers. Verify proper route propagation with the show ip bgp vpnv4 all ip-prejix
(IOS and lOS XE) or show bgp vpnv4 unicast ip-prejix (IOS XR) command on the
remote PE router.
Step 4 The VPNv4 routes that are received by the PE router need to be inserted into the
proper VRF. This insertion can be verified with the show ip route vrf (IOS and lOS
XE) or show route vrf (IOS XR) command. The validity of the import RTs can be
verified with the show ip bgp vpnv4 all ip-prejix (IOS and lOS XE) or show bgp
vpnv4 unicast ip-prejix (IOS XR) command, which displays the RTs attached to a
VPNv4 route. You can also verify the validity of the import RTs with the show ip
vrf detail (IOS and lOS XE) or show vrf detail (IOS XR) command, which lists the
import RTs for a VRF. At least one RT attached to the VPNv4 route needs to match
at least one RT in the VRF.

2-96 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Step 5 Next, the BGP routes that are received via MP-BGP and inserted into the VRF need
to be redistributed into the PE-CE routing protocol.
Step 6 Finally, the routes that are redistributed into the PE-CE routing protocol need to be
propagated to CE routers .

...--..
Is the Cisco Express
Forwarding entry correct on
r---_--l.-----,the ing ress PE ro ute r?

C2011Ci'""".rrlI>'it. .ffili*'".A1lrigi'ts......".......

After you have verified proper route exchange, start MPLS VPN data flow troubleshooting
using the checks that are listed in the next figures.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-97

 show cef vrf vrf-name ip-prefix/length detail

CE-Spoke CE-Spoke

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

One of the most common configuration mistakes related to data flow is the failure to enable
Cisco Express Forwarding on the ingress PE router interface. The presence of Cisco Express
Forwarding can be verified with the show cef interface command on Cisco lOS, lOS XE, and
lOS XR devices.
If Cisco Express Forwarding switching is enabled on the ingress interface, you can verify the
validity of the Cisco Express Forwarding entry and the associated label stack with the show ip
cef vrf vr.f-name ip-prefix detail (lOS and lOS XE) or show cef vrf vrf-name ip-prefix detail
(lOS XR) command. The top label in the stack should correspond to the BGP next-hop label as
displayed by the show mpls forwarding-table (lOS and lOS XE) or show mpls forwarding
(lOS XR) command on the ingress router. The second label in the stack should correspond to
the label allocated by the egress router. You can verify this label by using the show mpls
forwarding-table (lOS and lOS XE) or show mpls forwarding vrf (lOS XR) command on
the egress router.

2-98 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ..-.. -.-~--
Check for summarization issues. The BGP next hop should be
reachable as a host route.
Quick check-lfTTL propagation is disabled, the trace from PE-2 to
PE-1 should contain only one hop.
If needed, check LFIB values hop by hop.
Check for MTU issues on the path. MPLS VPN requires a larger label
header than pure MPLS.

CE-Spoke CE-Spoke

C2011Ci'""".r'lIi'<rit. .ffili*'".A1lrigi'ts......".......

If Cisco Express Forwarding is enabled on the ingress interface and the Cisco Express
Forwarding entry contains the proper labels, the data flow problem might lie inside the MPLS
core. Two common mistakes include summarization of BGP next hops inside the core IGP and
MTU issues.
The quickest way to diagnose summarization problems is to disable IP Time to Live (TTL)
propagation into the MPLS label header using the no mpls ip ttl-propagate (IOS and lOS XE)
or mpls ip ttl-propagate disable (IOS XR) configuration command on the provider router (P
router) and PE routers. The traceroute command from the ingress PE router toward the BGP
next hop should display no intermediate hops when TTL propagation is disabled. If
intermediate hops are displayed, the label-switched path (LSP) tunnel between PE routers is
broken at those hops, and the VPN traffic cannot flow.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-99

 show cef vrf vrf-name ip-prefix/length detail
show mpls forwarding vrf vrf-name value detail

CE-Spoke CE-Spoke

C2011Ci'""".rrl'<>'it. .ffili*'".A1lrigi'tsreoerv.....

As a last troubleshooting measure (usually not needed), you can verify the contents ofthe label
forwarding information base (LFIB) on the egress PE router and compare them with the second
label in the label stack on the ingress PE router. A mismatch indicates an internal Cisco lOS
Software error that you will need to report to the Cisco Technical Assistance Center (TAC).

Cisco lOS and lOS XE Cisco lOS XR

show ip ospf database show ospf database
showip bgp showbgp
show ip eigrp topology showeigrp topology

Ishow ip route
show mpls Idp bindings show mpls Idp bindings

show ip cef showcef

show ip cef vrf showcefvrf

show mpls forwarding-table

show mpls forwarding
show mpls forwarding-table vrf
show mpls forwarding vrf

C2011Ci""".r'lIi'<rit. .ffili*'".A1lrigi'tsreoerv.....

The figure shows the relevant commands for Cisco lOS, lOS XE, and Cisco lOS XR devices to
troubleshoot the control plane and the data plane of an MPLS router.

2-100 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary points that were discussed in this lesson.

2 3
OSPF as a PE-CE routing protocol is implemented as a separate routing
process.
BGP is very scalable and predictable as a PE-CE routing protocol.
MPLS VPN troubleshooting has two main steps: verifying routing
information flow and verifying proper data flow.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-101

2-102 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 41

Deploying IPv6 and MPLS

Overview
When deploying IPv6, you need to understand how to deploy it over a Multiprotocol Label
Switching (MPLS) network. This lesson introduces some concepts that can be used when you
design or deploy IPv6 in your environment.

Objectives
Upon completing this lesson, you will be able to describe how to configure routing protocols
between PE and CE routers. You will be able to meet this objective:
Describe the various methods that are used to deploy IPv6 over MPLS
Support for IPv6 in MPLS
This topic describes the various methods that are used to deploy IPv6 over MPLS.

There are several methods to provision IPv6 over an MPLS

network:
IPv6 tunnels configured on the CE
Layer 2 MPLS VPN
Cisco 6PE

C2012Ci:ocotlrrl'oritstrffili*'".A1lrigi'ts"'""'v.....

IPv6 over MPLS backbones enables isolated IPv6 domains to communicate with each other
over an MPLS IPv4 network. Depending on the deployment scheme that is chosen and on the
current MPLS implementation, enabling IPv6 services might require fewer backbone
infrastructure upgrades and less reconfiguration of core routers than enabling IPv4 services,
because forwarding is based on labels rather than on the IP header, thus providing a very cost-
effective strategy for the deployment of IPv6.
Additionally, the inherent VPN and traffic engineering services that are available within an
MPLS environment allow IPv6 networks to be combined into VPNs or extranets over an
infrastructure that supports IPv4 VPNs and MPLS terminal equipment. A variety of
deployment strategies are available or under development:
Deploying IPv6 by using tunnels on the customer edge (CE) routers: This strategy has
no impact on and requires no changes to the MPLS provider (P) routers or provider edge
(PE) routers. The strategy uses IPv4 tunnels that are terminated on the CE routers to
encapsulate the IPv6 traffic, which then appears as IPv4 traffic within the MPLS provider
network.
Deploying IPv6 over a Layer 2 MPLS VPN: This strategy, applicable only on specific
Cisco routers (such as the Cisco 12000 and 7600 Series Routers), also requires no changes
to the core routing mechanisms, assuming that the provider is already capable of supporting
Layer 2 MPLS VPNs.
Deploying IPv6 on the PE routers via Cisco IPv6 Provider Edge Router over MPLS
(6PE): This strategy requires changes to the PE routers to support a dual-stack
implementation, but all the core functions remain IPv4-only, and so the P routers require no
changes.

2-104 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 IPv6 IPv6
IPv6-in-IPv4
Tunnels

Dual-Stack
IPv4-1Pv6
CE Routers
PE

IPv4
Dual-Stack
IPv4-IPv6
CE Routers

IPv4

C2012Ci'""""rd'oritstrffili*'".AIlrigi'ts......".......

Using tunnels on the CE routers is the simplest way to deploy IPv6 over MPLS networks. This
method has no impact on the operation or infrastructure ofMPLS and requires no changes to
the P routers in the core or to the PE routers that connect to customers.
IPv6 tunnels that are configured on a CE have these characteristics:
The CE routers are IPv6-aware (dual stack).
A mesh of IPv6-over-IPv4 tunnels connects CE to CEo
Overhead includes the IPv4 header and MPLS header.
MPLS and VPN support native IPv4 and IPv6 tunnels.
Communication between the remote IPv6 domains uses standard tunneling mechanisms,
running IPv6 over IPv4 tunnels, as the way that MPLS VPNs support native IPv4 tunnels. The
CE routers need to be upgraded to dual stack and configured by using manually configured
tunnels or 6t04 tunnels. However, communication with the PE routers is IPv4-only, and the
traffic appears to be IPv4 to the MPLS domain. The dual-stack routers use the 6t04 tunnel
addresses or an IPv6 prefix that is assigned by a virtual IPv6 ISP (possibly other than the MPLS
provider). The figure shows an example for deployment ofIPv6 by using tunnels on the CE
routers.
This alternative is attractive because it has no impact on the MPLS P or PE routers, because it
uses IPv4 tunnels to encapsulate IPv6 traffic. After being encapsulated, the traffic appears as
IPv4 traffic within the network. However, this option does not permit the delegation of a /48
prefix address from the MPLS service provider address space to customers. Nonetheless, it is
probably useful for specific applications, such as adding IPv6 services to an MPLS VPN by
configuring the tunnel on the CEo All Cisco lOS Software tunneling mechanisms support this
capability.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-105

Pros
From the provider standpoint, using tunnels on CE routers is the simplest way to deploy IPv6
over MPLS networks. This approach has no impact on the operation or infrastructure of MPLS
and requires no changes to either the P routers in the core or the PE routers that connect to
customers.

Cons
Using tunnels on CE routers presents the same problems as any configured tunnel scheme. As
the number of tunnel partners grows, the number of tunnels in an any-to-any mesh increases
quickly, making management of the tunnels error-prone and expensive. This problem is not
specific to MPLS but rather is a general issue with manually configured tunnels. Automatic
tunnels (such as 6t04) can be used to reduce the effort that is required to manage tunnels on the
CE, but automatic tunnels have other issues that must be considered.
Delegating a global IPv6 prefix for an ISP is also difficult because the ISP does not furnish the
IPv6 connectivity; it merely provides transport. Even when the ISP does allocate an address,
early traffic levels will be modest, and later traffic might shift from IPv4 to IPv6. Therefore,
there is not necessarily a net increase in carried packets. Because the service provider provides
transport only over IPv4 over MPLS, this scenario offers little possibility of generating
additional revenue for the provider.
From the provider standpoint, using tunnels on CE routers is not a good mechanism to carry
IPv6 traffic across an IPv4-based MPLS network. The provider basically puts responsibility for
IPv6 transit on the end customers, because the MPLS provider recognizes only IPv4 traffic.

2-106 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

Using any circuit transport to deploy IPv6 over MPLS networks has no impact on the operation
or infrastructure of MPLS. Circuit transport requires no changes to either the P routers in the
core or the PE routers that connect to customers, assuming that the provider has already
implemented circuit-over-MPLS technology. This design is a Layer 2 solution.
IPv6 using a Layer 2 MPLS VPN has these characteristics:
IPv6 is run over pseudowires (see the IETF document draft-ietf-pwe3-wildcard-pw-type-
02.txt).
Edge MPLS routers need to support a Layer 2 MPLS VPN.
A mesh of virtual circuits connects PE to PE.
The PE routers are regular IPv4 routers that are used to aggregate customer IPv6 routers.
Communication between remote IPv6 domains runs native IPv6 protocols over a dedicated
link, in which the underlying mechanisms are fully transparent to IPv6. The IPv6 traffic is
tunneled by using a Layer 2 MPLS VPN or Ethernet over MPLS (EoMPLS), with the IPv6
routers connecting through an ATM or Ethernet interface, respectively. The PE will require an
upgrade to a Layer 2 MPLS VPN if it does not already contain that support. The figure shows
an example ofIPv6 deployment over any circuit transport over MPLS.
The alternative of IPv6 over Layer 2 MPLS VPN is likely to be used by service providers that
have ATM or Ethernet links to CE routers. This alternative is fully transparent to users.
However, like the previous option, this option does not permit the delegation of a /48 prefix
address from the service provider address space to customers. In addition, IPv6 over Layer 2
MPLS VPN does not scale as well as 6PE because service providers need to create circuits
through their MPLS networks for their customers.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-107

Pros
There are some advantages to using IPv6 over a Layer 2 MPLS VPN:
Simple solution for service providers with ATM or Ethernet links to CE routers and circuit
over MPLS capability
Fully transparent IPv6 communication, from the provider perspective
No required reconfiguration of the MPLS core

Cons
There are also some disadvantages to using IPv6 over a Layer 2 MPLS VPN:
Equivalent to Layer 2 VPN complexity
The need to upgrade PE routers that connect to customers to a Layer 2 MPLS VPN
The difficulty of delegating a global IPv6 prefix for an ISP

Description of Pros and Cons

Additional points about IPv6 over Layer 2 MPLS VPN are as follows:
The use of a Layer 2 MPLS VPN to deploy IPv6 over MPLS networks does not affect the
operation or infrastructure ofMPLS. Neither the P router software version nor the
configuration is changed. However, PE routers that connect to customers must be upgraded
to Layer 2 MPLS VPN. Delegating a global IPv6 prefix for an ISP is also difficult.
Communication between remote IPv6 domains runs native IPv6 protocols over a dedicated
link in which the underlying mechanisms are fully transparent to IPv6. The IPv6 traffic is
tunneled via a Layer 2 MPLS VPN, with the IPv6 routers connected through an ATM or
Ethernet interface. This potentially heavy tunnel meshing is a drawback of the solution.

2-108 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 MP-16GP
2001 :D68:0002:: IPv6
Sessions

P
6PE

/1
I / I
I / I
~ I / I
192.0.2.0 IPv4
1/ I
,. I
2001 :D68:0003:: / I
I I
/ I I
P ,/ I I

C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

Cisco 6PE enables IPv6 islands to communicate with each other over an MPLS IPv4 core
network by using MPLS label-switched paths (LSPs). The method relies on Border Gateway
Protocol (BGP) extensions in the IPv4 network PE routers (Cisco 6PE routers) to exchange
IPv6 reachability information and an MPLS label for each IPv6 address prefix that is
announced. Cisco 6PE routers are dual stack (IPv4 and IPv6). This design is a Layer 3 solution.
For the IPv6 transport to be transparent to all but Cisco 6PE routers, you must impose a
hierarchy of labels at the Cisco 6PE ingress router. The top label provides connectivity inside
the IPv4 MPLS core network. Label Description Protocol (LDP), Tag Distribution Protocol
(TDP), or Resource Reservation Protocol (RSVP) distributes this label. The inner label is used
for IPv6 forwarding at each Cisco 6PE egress router. This label is distributed by Multiprotocol
Border Gateway Protocol (MP-BGP) in the VPN-IPv6 address family.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-109

 2001 :D68:0002:: IPv6 MP-16GP
Sessions

P :

::=====eI jD.-=-. . -=-~-=

6 PE 6PE

/ I
I I / I
I I / I
I I / I
I 1/ I
I ,. I
2001 :D68:0003:: I / I I
I / I
I I
I P ,/ I I

IPv6-U naware
192.0.2.30
No Core Upgrade

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

Cisco 6PE uses MPLS to minimize changes in the ISP core network.
With Cisco 6PE, IPv6 forwarding in the core of the MPLS network is done by label switching,
eliminating the need for either IPv6-over-IPv4 tunnels or an additional Layer 2 encapsulation.
It allows the appearance of a native IPv6 service to be offered across the network.
Each PE router that must support IPv6 connectivity needs to be upgraded to dual stack
(becoming a Cisco 6PE router) and configured to run MPLS on the interfaces that connect to
the core. Depending on the site requirements, each router can be configured to forward IPv6 or
IPv6 and IPv4 traffic on the interfaces to the CE routers, thus providing the ability to offer only
native IPv6 or both IPv6 and native IPv4 services. The Cisco 6PE router exchanges either IPv4
or IPv6 routing information with the CE router through any of the supported routing protocols
(depending on the connection). It switches IPv4 and IPv6 traffic over the native IPv4 and IPv6
interfaces that do not run MPLS.
The Cisco 6PE router uses MP-BGP to exchange reachability information with the other 6PE
routers in the MPLS domain. It shares a common IPv4 routing protocol (such as Open Shortest
Path First [OSPF] or Intermediate System-to-Intermediate System [IS-IS]) with the other P and
PE devices in the domain. Interior gateway protocol (IGP) is used for deriving next-hop
information.
The Cisco 6PE routers encapsulate IPv6 traffic by using two levels ofMPLS labels. The top
label is distributed by an LDP or TDP that devices in the core use to carry the packet to the
egress Cisco 6PE router, using IPv4 routing information. The second or bottom label is
associated with the IPv6 prefix ofthe destination through MP-BGP version 4 (MP-BGP4).

Multiprotocol Extensions for BGP4

Multiprotocol extensions for BGP4 are characterized as follows:
BGP4 is the primary mechanism to exchange IPv6 reachability information for Cisco 6PE
and Cisco IPv6 VPN Provider Edge Router over MPLS (6VPE).
Label binding information is piggybacked with the prefix advertisement in a Network
Layer Reachability Information (NLRI) attribute.

2-110 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 BGP protocol extensions, which carry routing information about other protocols, include
multicast, MPLS, IPv6, VPN-IPv4, labeled IPv6 unicast (Cisco 6PE), and Cisco 6VPE.
The exchange of multiprotocol NLRI must be negotiated at the session setup. The exchange
uses BGP capability advertisement negotiation procedures.
MP-BGP is the supported exterior gateway protocol (EGP) for IPv6. MP-BGP extensions for
IPv6 support the same features and functionality as IPv4 BGP. IPv6 enhancements to MP-BGP
include support for an IPv6 address family, NLRI, and next-hop attributes that use IPv6
addresses.
The Cisco 6PE multipath feature uses Multiprotocol Internal BGP (MP-IBGP) to distribute
IPv6 routes over the MPLS IPv4 core network and to attach an MPLS label to each route.

Note See RFC 2858, Multiprotocol Extensions for BGP4 for more information. (RFC 2283 is
obsolete.)

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-111

 2001:DB8:4::
G)
1------------------, CE2
~ :------~-:.-:.-:.-:.?
Loopback 0: -r___....L..- Loopback 0:
~ ~ ~ ~ ~
192.0.2.1
6PE1
:
'2'
\6J 6PE2
192.0.2.133

Translation of IPv6 BGP

Next Hop into IPv4Addresses

Recursion of this Address

via IGPv4

... _ _ _ _ _ 1. IGPv4 advertises the reachability of 192.0.2.133.

2. LOPv4 binds a label to 192.0.2.133.

..c-._.-> 3. MP-BGP advertises 2001 :OB8:4:: and binds a (second-level) label. The
IPv6 next hop is an IPv4-mapped IPv6 address built from 192.0.2.133.

4. The BGP next hop for the destination IPv6 prefix is set to an
IPv4-compatible address, which is built from the 6PE2 router.

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

This example describes how IPv6 transport is achieved across an IPv4-based MPLS provider
network that is running an IPv4 IGP such as OSPF version 2 (OSPFv2), and how reachability is
established between Cisco 6PEs.
1. IGPv4 advertises the reachability of PE routers.

The Cisco 6PE routers exchange IPv6 routing information through MP-BGP. The Cisco 6PE
routers peer together through MP-BGP4 to exchange IPv6 reachability with the MPLS cloud
and to perform IPv6 forwarding. IGP advertises the reachability of the BGP next-hop address;
that is, 192.0.2.133.
2. LDPv4 binds labels to create LSPs.

3. The 6PE2 router sends MP-BGP advertisements to the ingress PE router, 6PE1.

4. The BGP next hop for the destination IPv6 prefix is set to an IPv4-compatible address,
which is built from the 6PE2 router.

The MPLS label binds to the IPv4- or IPv6-mapped address, which is formed from the IPv4
address of 6PE2 (192.0.2.133). Thus, other Cisco 6PE routers (which are part of the mesh
edges around the MPLS core) see 6PE2: :FFFF: 192.0.2.133 as the next-hop IPv6 address when
a packet needs to be forwarded to the 200l:DB8:4:: network.

2-112 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEOGE) v1.0 2012 Cisco Systems, Inc.
 MPLS label POP and IPv6 forwarding:
6PE2 receives an MPLS packet.
Lookup is done on label.
The result is POP label and IPv6100kup on Ihe
IPv6 destination.
2001:068:0002::
IPv6 Packet 10
2001 :068:4::

LOP/IGPv4 MP-6GP Label

Label1106PE2 102001:068:4:: I..===~

LOP/IGPv4 MP-6GP Label

Label1 to 6PE2 t02001:068:4:: ......""""......""-'

C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

This example describes how routers forward packets in data plane:

1. An IPv6 packet arrives at the ingress Cisco 6PE router.

An IPv6 packet that is destined to a network with prefix 2001 :DB8:4:: enters through 6PEl.
The MPLS core must be used to carry the packet to its destination, via 6PE2. The routing
configuration via IBGP has already advertised the next-hop IPv6 address (::FFFF: 192.0.2.133)
to 6PEl.
2. The ingress Cisco 6PE router binds MP-BGP (inner) and LDP-IGPv4 (top, or outer) labels
to the packet.

6PEl does a lookup on the IPv6 prefix. Two labels are then created and bound to the packet.
MP-BGP binds 2001 :DB8:4:: and LDP-IGPv4 binds its label to the IPv4 address of 6PE2
(192.0.2.133). In this way, MP-BGP can carry either IPv4 packets or IPv6 packets from one
Cisco 6PE router to another.
3. The IPv6 packet is forwarded to the first P router.

MPLS label switching is IPv6-unaware. When PI receives the MPLS packet, it looks only at
the LDP-distributed MPLS label, and a lookup is performed. The result is the imposition of
Label2, and the MPLS packet is forwarded to P2. The IPv6 traffic is transparent to the core of
the MPLS network.
4. The IPv6 packet is forwarded to the next P router.

P2 receives an MPLS packet, which performs an MPLS lookup on Label2 and results in the
point of presence (POP) label. As a result, the packet is forwarded to the IPv4 address to which
it is bound. The packet still has the inner MPLS label, which is required because a raw IPv6
packet is not allowed within the MPLS core.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-113

 5. The egress PE routers remove the MPLS labels and forward the IPv6 packet.

The 6PE2 router examines the MPLS packet. The result is that the remaining MPLS label is
removed and an IPv6 lookup on the IPv6 destination address is performed, using normal IP
destination-based routing. The Cisco 6PE router forwards the IPv6 packet on to the CE device,
based on the results of the IPv6 lookup on the IPv6 destination address.

2-114 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Full Mesh of
MP-IBGP -------. IPv6
Sessions

IPv4 IPv6 IGP MP-BGP IPv4

C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

Using Cisco 6PE for IPv6 integration has these benefits:

A IPv6 CE has only one PE routing peer, regardless of how many remote IPv6 CEs it
communicates with.
No changes are required on an IPv6 CE when remote CEs are added or removed.
Reachability is automatically learned via MP-BGP.
No tunnel or circuit configuration is required.
Cisco 6PE offers a scalable and flexible solution. Benefits are analogous to the Layer 3
VPN solution for IPv4, in RFC 2547bis.
Service providers that already deploy MPLS, or plan to do so, can obtain these benefits from
Cisco 6PE:
There is minimal operational cost and risk and no impact on existing IPv4 and MPLS
service, because only PE routers are upgraded. A Cisco 6PE router can be an existing PE
router or a new router that is dedicated to IPv6 traffic.
There is no impact on IPv6 CE routers. The ISP can connect to any CE device that runs
static, IGP, or EGP ready-for-production services. An ISP can delegate IPv6 prefixes to
customers.
Nondisruptive IPv6 can be introduced into an existing MPLS service Cisco 6PE router at
any time.
There are some drawbacks to using Cisco 6PE for integration:
This option makes sense only when the network already runs MPLS.
It requires knowledge of MPLS and BGP technologies.
It requires dual stack and software upgrades on existing PE routers or deployment of
dedicated Cisco 6PE routers.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-115

 interface GigabitEthernet O/OjO/O
description Customer
ipv6 enable
ipv6 address 2001:0db8::14j96

interface GigabitEthernet 0/0/0/1

description CORE
ipv4 address 192.168.2.2 255.255.255.252
!
mpl. ldp
interface GigabitEthernet 0/0/0/1
router bgp 64500
neighbor 192.168.2.1
remote-as 64500
address-family ipv4 unicast
address-family ipv6 labeled-unicast

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

Follow these steps to deploy Cisco 6PE:

Step 1 Configure Cisco 6PE routers as dual-stack routers.
Step 2 Configure Cisco 6PE routers to be part of the IPv4 network and configure MPLS.
Step 3 Specify the interface from which locally generated packets take their source IPv6
addresses.
Step 4 Bind and advertise aggregate labels.

Example
In the network that is shown in the figure, two configuration tasks are required at the 6PEI
router to enable the Cisco 6PE feature.
The CE router, CEI, is configured to forward its IPv6 traffic to the 6PEI router. The PI router
in the core of the network is assumed to be running MPLS, a label distribution protocol, an
IPv4 IGP, and Cisco Express Forwarding or distributed Cisco Express Forwarding. PI does not
require any new configuration to enable the Cisco 6PE feature. New configuration tasks are
also not required for the CE I router.
The Cisco 6PE routers, 6PEI and 6PE2, must be members of the core IPv4 network. The Cisco
6PE router interfaces that are attached to the core network must run MPLS, the same LDP, and
the same IPv4 IGP as in the core network.
The 6PE routers must also be configured to be dual stack to run both IPv4 and IPv6.

2-116 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Restrictions
These restrictions apply when implementing the Cisco 6PE feature:
Core MPLS routers support MPLS and IPv4 only, so they cannot forward or create any
IPv6 Internet Control Message Protocol (ICMP) messages.
Cisco 6PE does not provide load balancing between an MPLS path and an IPv6 path. If
both are available, the MPLS path is always preferred. Load balancing between two MPLS
paths is possible.
When MP-IBGP multipath is enabled on the Cisco 6PE router, all labeled paths are
installed in the forwarding table with MPLS information (label stack) when MPLS
available. This functionality enables Cisco 6PE to perform load balancing.

Can be compared to a regular IPv4 MPLS VPN PE, with the additon of
IPv6 support within a VRF
Provides logically separate routing table entries for VPN member
devices
Can be deployed over an existing IPv4 backbone
Does not require modification of the MPLS core routers

C2012Ci'""""ndIoritstrffili*'".A1lrigi'ts......,".....

IPv6 MPLS VPNs are like IPv4 MPLS VPNs. Customer sites are connected over the shared
MPLS backbone so that customers have reachability to their own VPN partner sites only. As
with IPv4 MPLS VPNs, this outsourced intranet remains private and is seen by the customer as
a closed network that runs on a dedicated network.
The inherent VPN and terminal equipment services that are available within an MPLS
environment allow IPv6 networks to be combined into VPNs or extranets over an infrastructure
that supports IPv4 VPNs and MPLS terminal equipment.
ISPs that offer MPLS VPNs for IPv4 can use Cisco 6VPE to add IPv6 services. This approach
has these benefits:
There is no modification on the MPLS core.
Cisco 6VPE supports both IPv4 and IPv6 VPNs concurrently on the same interfaces.
Configuration and operation ofIPv6 VPNs are the same as for IPv4 VPNs.

2012 Cisco Systems, Inc. MPLSLayer3VPNs 2-117

 Customer A Customer A
Full Mesh of
2001 :OB8:422:: MP-IBGP
Sessions
P
6VPE

I r
\ I I
\ I I
\ I
\ I
/ I
2001 :OB8:429::
I \ I
I \ I
I \ I
P , \ I

2001 :OB8:824::

CustomerB CustomerB

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

Service providers that offer MPLS IPv4 VPN services to their customers might look forward to
adding IPv6 VPN services to their portfolios. A VPN is said to be an IPv6 VPN when a CE
router turns on native IPv6 over an interface or subinterface to the PE router. Cisco 6VPE adds
IPv6 VPN capability to Cisco 6PE and enables an ISP to deliver services that are similar to
IPv4.
As with IPv4 VPN route distribution, BGP and its extensions are used to distribute routes from
an IPv6 VPN site to all other Cisco 6VPE routers that connect to a site of the same IPv6 VPN.
Cisco 6PE routers use virtual routing and forwarding (VRF) tables to separately maintain the
reachability and forwarding information of each IPv6 VPN (as with IPv4 VPN).
The IETF standard defines a specification that lets a service provider use an MPLS-enabled
IPv4 backbone to provide VPNs for its IPv6 customers. The standard method is based on VPNs
as described in RFC 4364, BGPIMPLS IP VPNs. In a BGP/MPLS VPN, MPLS is used to
forward packets over the backbone, and Multiprotocol Border Gateway Protocol (MP-BGP) is
used to distribute VPN routes over the service provider backbone.
MP-BGP allows generic operations over both IPv4 and IPv6. For example, MP-BGP defines an
IPv6 VPN address family and describes the route distribution and MPLS tunnel selection.

2-118 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customer A Customer A
Full Mesh of
2001 :OB8:422:: MP-IBGP
Sessions

L..-_ _ 8=;=8--:ClVIi~ED:2====:
~6V:EP~E1 =
P : P

I I 'I
I I , I
I I, I
I I, I
Customer A
2001 :OB8:429:: I ," I I
I , I I
, I I
I I I

2001 :OB8:824::

CustomerB
interface EthernetO
ip vrf forwarding CustomerA
ipv6 address 2001:db8:101::1j64

C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

Forwarding in Cisco 6VPE follows these steps:

1. The ingress PE router receives an IPv6 packet and attaches an MPLS label.

When a Cisco 6VPE router receives an IPv6 packet from CEl, the router looks up the IPv6
packet destination address in the VRF table. This address enables 6VPEl to find a VPN IPv6
route. The VPN IPv6 router has an associated MPLS label and an associated BGP next hop.
This MPLS label is imposed on the IPv6 packet.
2. The ingress PE router attaches a second label to the IPv6 packet.

6VPEl pushes another label, which is bound by LDP IGP to the IPv4 address of the BGP next
hop, to reach 6VPE2 through the MPLS cloud on the label stack of the labeled IPv6 VPN
packet. This top-most imposed label corresponds to the LSP that starts on 6VPEl and ends on
6VPE2. The bottom label is bound to the IPv6 VPN prefix via BGP. All the P routers in the
backbone network switch the VPN packet, based only on the top label in the stack, which
points toward the 6VPE2 router. Because of the normal MPLS forwarding rules, the P routers
never look beyond the first label and are thus completely unaware of the second label or the
IPv6 VPN packet that is carried across the backbone network.
3. The egress router receives the labeled IPv6 packet, removes the labels, and forwards the
packet to the CE router.

The egress PE router, 6VPE2, receives the labeled IPv6 VPN packet, drops the first label, and
performs a lookup on the second label, which uniquely identifies the target VRF table and
sometimes even the outgoing interface on 6VPE2. A lookup is performed in the target VRF
table, and the IPv6 packet is sent toward the proper CE router in the IPv6 domain or site.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-119

 The configuration steps for 6VPE on PE routers are as follows:
Enable IPv6 routing and IPv6 Cisco Express Forwarding.
Configure MP-BGP for VPN route exchange.
Configure the customer IPv6 VRF tables.
Configure the customerVRF interfaces.
Configure the customerVRF (PE-CE) IPv6 routing protocols or static
IPv6 routes.
Redistribute the CE-PE routing protocol VRF routes into MP-BGP.

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

To successfully implement 6VPE, you must apply all the steps that the figure shows.
This task assumes that these steps have already been completed:
A loopback interface is configured.
LDP is configured.
MPLS is enabled on interfaces.
The backbone IGP (OSPF or IS-IS) is configured.

2-120 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Cisco lOS and Cisco lOS
10SXE XR
vrf Customer A
address-family ipv6 unicast
import route-target
64500,1

export route-target
vrf Cus tomer A 64500,1
rd 64500,1
route-target export 64500: 1 router bgp 64500
route-target import 64500: 1 address-family vpnv6 unicast
! !
router bgp 64512 neighbor 10.1.1.1
neighbor 10.1.1. 4 remote-as 64500 remote-as 64500
neighbor 10.1.1.4 update-source LoopbackO update-source LoopbackO
! !
address-family vpnv6 address-family vpnv6 unicast
neighbor 10.1.1.4 activate !
neighbor 10.1.1.4 send-community both vrf Customer A
! rd 64500,1
address-family ipv6 vrf Customer A address-family ipv6 unicast
redistribute connected redistribute connected

ipv6 route vrf Customer_A router static

2003:430:210:1: :/64 GigabitBthernetO/O vrf Customer A
address-family ipv6 unicast
2b11: :2f01:4c GigabitBthernetO/O/O/O

C2012Ci'""""rd'oritstrffili*'".A1lrigi'ts......".......

The figure shows a sample configuration ofPE routers. Router 6PEI is running Cisco lOS and
IOS XE Software, while 6PE2 is running Cisco lOS XR Software. Both PE routers have an
MP-BGP session for route exchange and a configured VPN (Customer_A).

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-121

 router#
Ishow ipv6 route vrf vrf-name

Displays the IPv6 routing table associated with a VPN VRF instance

router#
I show bgp vpnv6 unicast vrf vrf-name

Displays VPN entries in a BGP table

router#
Ishow running-config vrf vrf-name
Displays VPN VRF configuration

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts"'""'v.....

The figure shows commonly used commands for configuration verification.

6PE#show ipv6 route vrf VPN_A
IPv6 Routing Table - VPN_A - 6 entries
Codes: C - Connected, L - Local, S - Static, U - Per-user Static route
B - BGP, R - RIP, 11 - ISIS L1, 12 - ISIS L2
IA - ISIS interarea, IS - ISIS summary
o - OSPF Intra, 01 - OSPF Inter, OE1 - OSPF ext 1, OE2 - OSPF ext 2
ON1 - OSPF NSSA ext 1, ON2 - OSPF NSSA ext 2
C 2003: 410: 210: 1: : /64 [0/0]
via Serial2/0, directly connected
L 2003:410:210:1:2DO:63FF:FE54:7000/128 [0/0]
via Serial2/0, receive
B 2003: 420: 210: 1: : /64 [200/0] (line 1)
via 10.1.1.4%Default-IP-Routing-Table, indirectly connected
S 2003:430:210:1: :/64 [1/0] (line 2)
via Serial2/0, directly connected
B 2003:440:210:1: :/64 [200/0] (line 3)
via 10.1.1.4%Default-IP-Routing-Table, indirectly connected
L FFOO: : /8 [0/0]
via NullO, receive

2-122 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary point that was discussed in this lesson.

2 3
IPv6 and IPv6 VPNs are treated as another service that can be added
on the edge over the stable multiservice IPv4 MPLS core.

C2012Ci'""""rrl'oritstrffili*'".A1lrigi'ts......".......

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-123

2-124 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Summary
This topic summarizes the primary points that were discussed in this module.

3
To implement an MPLS Layer 3 VPN backbone, MPLS must be
configured in the core and edge network and MP-BGP must be
configured on the PE routers.
All routing protocols that support per-VRF routing can be used for route
exchange between PE and CE devices.
BGP is very scalable and predictable as a PE-CE routing protocol.
IPv6 deployment strategies include IPv6 tunnels that are configured on
the CE, IPv6 over Layer2 MPLS VPN, Cisco 6PE, and native MPLS
support of IPv6.

C2012Ci'""""rrl'oritstrflili*'".A1lrigi'ts......".......

A Multiprotocol Label Switching (MPLS) VPN implementation involves virtual routing and
forwarding (VRF) tables, the interaction between customer edge (CE)-to-provider (P) routing
protocols, and Multiprotocol Border Gateway Protocol (MP-BGP) in the service provider
backbone.

References
For additional information, refer to these resources:
Cisco Systems, Inc. "Configuring MPLS Layer 3 VPNs."
http://www.cisco.com/en/US/docslios/mpls/configuration/guide/mp_cfg_Iayer3_vpn.html.
Cisco Systems, Inc. "Implementing MPLS Layer 3 VPNs on Cisco lOS XR Software."
http://www.cisco.com/en/US/docslios_xr_swliosxrJ3. 7/mpls/configuration/guide/gc3 7v3.
html.
Cisco Systems, Inc. The "MPLS VPN Implementation" module in the Implementing Cisco
MPLS (MPLS) v2.3 course.
Cisco Systems, Inc. The "Identifying IPv6 Service Provider Deployment" module in the
IPv6 Fundamentals, Design, and Deployment (IP6FD) course.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-125

2-126 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Self-Check
Use the questions here to review what you learned in this module. The correct answers and
solutions are found in the Module Self-Check Answer Key.

Q I) In an MPLS VPN implementation, what is a VRF? (Source: Implementing MPLS

Layer 3 VPN Backbones)
A) the routing and forwarding instance for all sites belonging to a single customer
B) the routing and forwarding instance for all sites belonging to a single customer
location
C) the routing and forwarding instance for all sites using a common routing
protocol
D) the routing and forwarding instance for a set of sites with identical connectivity
requirements
Q2) Which two protocols are VPN-aware? (Choose two.) (Source: Implementing MPLS
Layer 3 VPN Backbones)
A) RIPv2
B) IS-IS
C) ODR
D) EIGRP
E) RIPvl

Q3) VRFs are assigned to an interface. (Source: Implementing MPLS Layer 3 VPN
Backbones)
A) true
B) false
Q4) Which command do you use to create a VRF named "VPNA" on Cisco lOS XR
devices? (Source: Implementing MPLS Layer 3 VPN Backbones)
A) ip vrfVPNA
B) vrfVPNA
C) ip vrf forwarding VPNA
D) vrf forwarding VPNA

Q5) Which command do you use to associate interface Gigabit Ethernet 0/0/0/0 with a VRF
named "VPNA" on Cisco lOS XR devices? (Source: Implementing MPLS Layer 3
VPN Backbones)
A) ip vrfVPNA
B) ip vrf VPNA interface GigabitEthernetO/O/O/O
C) ip vrf forwarding VPNA
D) vrfVPNA
E) vrf VPNA interface GigabitEthernetO/O/O/O

Q6) What happens to the existing IP address of an interface when you associate the
interface with a VRF? (Source: Implementing MPLS Layer 3 VPN Backbones)
A) It remains unchanged.
B) It is removed from the interface.
C) It is changed to the Loopback 0 address.
D) It is moved under the VRF configuration.

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-127

 Q7) Which command configures the redistribution of static VRF routes between PE routers
that are running Cisco lOS XR Software? (Source: Connecting Customers Using
Simple Routing Protocols)
A) RP IO/RPOICPUO :router(config)# redistribute static
B) RP IO/RPOICPUO:router( config-router)# redistribute static
C) RP IO/RPOICPUO:router( config-bgp)# redistribute static
D) RP/o/RPOICPUO:router( config-bgp-vrf)# redistribute static
E) RP IO/RPOICPUO :router(config-bgp-vrf-af)# redistribute static

Q8) How is each routing context implemented in OSPF? (Source: Connecting Customers
Using BGP or OSPF)
A) by redistributing into MP-BGP
B) as a separate routing process
C) by assigning it to an interface
D) as a separate isolated instance of the same routing protocol
Q9) In the SOO extended community attribute in MP-BGP, what does "SOO" stand for?
(Source: Connecting Customers Using BGP or OSPF)
A) Site of Origin
B) Sites of Origin
C) State of Origin
D) Source of Origin

QIO) Which command do you use to display the contents of the LFIB table on a PE router
running Cisco lOS XR Software? (Source: Connecting Customers Using BGP or
OSPF)
A) show mpls forwarding-table
B) show mpls forwarding-table vrf
C) show mpls forwarding
D) show mpls forwarding vrf

Q 11 ) No tunnel or circuit configuration is required when using Cisco 6PE for IPv6
integration. (Source: Deploying IPV6 and MPLS)
A) true
B) false

2-128 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Self-Check Answer Key
Ql) D
Q2) A,D
Q3) B
Q4) B
Q5) D
Q6) B
Q7) E
Q8) B
Q9) A
QlO) C
Qll) A

2012 Cisco Systems, Inc. MPLS Layer 3 VPNs 2-129

2-130 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Module 31

Complex MPLS Layer 3 VPNs

Overview
This module discusses some advanced configuration features that can help increase the stability
of the Multiprotocol Label Switching (MPLS) VPN backbone. The first lesson discusses
various MPLS VPN features that a service provider might use to help meet service
requirements, and looks at various types of VPN solutions and topologies.
Integrating Internet access with an MPLS VPN solution is one of the most common service
provider business requirements. The second lesson provides a good understanding of
underlying design issues, several potential design scenarios, and some sample configurations.
Various topologies and implementation methods are discussed, along with ways to separate
Internet access from VPN services.
Deployments of MPLS have become routine in large-scale global networks, which demand
solutions to complex business and network problems. The third lesson describes the two
primary components of the Cisco lOS MPLS Inter-Domain Solution: interautonomous system
(inter-AS) and Carrier Supporting Carrier (CSC).

Module Objectives
Upon completing this module, you will be able to describe and implement the most important
complex Layer 3 VPN features. You will be able to meet these objectives:
Describe the requirements, usage, and solutions that are associated with overlapping,
central services, and managed CE router service VPNs
Describe common customer Internet connectivity scenarios and identify design models for
combining Internet access with MPLS Layer 3 VPN services
Introduce CSC and inter-AS features with different implementation options
3-2 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 11

Implementing Complex MPLS

Layer 3 VPNs
Overview
This lesson discusses some advanced configuration features that can help increase the stability
of the Multiprotocol Label Switching (MPLS) VPN backbone. The lesson also discusses
various MPLS VPN features that a service provider might use to help meet service
requirements, and looks at various types of VPN solutions and topologies.

Objectives
Upon completing this lesson, you will be able to describe the requirements, usage, and
solutions associated with overlapping, central services, and managed CE router service VPNs.
You will be able to meet these objectives:
Describe overlapping VPNs
Describe central service VPNs and advanced VRF features
Describe managed CE router service
Overlapping VPNs
This topic describes how a service provider can configure overlapping VPNs. Overlapping
VPNs are usually used to connect parts of two separate VPNs.

Complex MPLS Layer 3 VPNs are part of the Cisco IP NGN

infrastructure layer.
Layer 3 VPNs are usually configured on IP edge devices.
MPLS runs on IP core devices.
C2012Ciscoamlloritsafflllllte!l.AII~ghtll"",erved.

Complex MPLS Layer 3 VPNs include the following traits:

Complex MPLS Layer 3 VPNs are part of the Cisco IP Next-Generation Network (NGN)
infrastructure layer
Layer 3 VPNs are usually configured on IP edge devices
MPLS runs on IP core devices

3-4 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 --~

<
Central sites communicate
with each other :>
Central sites are reachable from multiple VPNs:
- Overlapping VPN
IP addressing in common sites should not overlap:
- NAT can be used when networks overlap.

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

When two VPN customers want to share information, they might decide to interconnect their
central sites. To achieve this interconnection, two simple VPNs are created, each containing a
customer central site and its remote sites. Then a third VPN, which partially overlaps with the
customer VPNs but connects only their central sites, is created. The central sites can talk to
each other. Each central site can also talk to the remote sites in its simple VPN, but not to the
remote sites belonging to the other customer simple VPN. The addresses used in the central
sites, however, must be unique in both VPNs.
Another option is to use dual Network Address Translation (NAT) with a registered address to
be imported and exported between the two central sites.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-5

 At least one customersite needs to be reachable by sites in different
VPNs:
- A service provider may provide services to many customers.
- Some service provider customers may want connectivityto one of their
partners through the MPLS network.
- Limitvisibilitybetween different departments in an organization.

sp

CustomerA CustomerC

CustomerB

CustomerB

Typical uses for overlapping VPNs include the following:

A service provider may provide services to its customers.
Companies that use MPLS VPNs to implement both intranet and extranet services might
use overlapping VPNs. In this scenario, each company participating in the extranet VPN
would probably deploy a security mechanism on its customer edge (CE) routers to prevent
other companies that are participating in the VPN from gaining access to other sites in its
VPN.
A security-conscious company might decide to limit visibility between different
departments in the organization. Overlapping VPNs might be used as a solution.

Note Security issues might force an enterprise network to be migrated to an MPLS VPN even if it
is not using MPLS VPN services from a service provider.

3-6 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Import
Export
RT 1:1000

Customer A (central) import and export:

- RT 1:210 (customerVPN)
- RT 1:1000 (overlapping VPN)
Customer B (central) import and export:
- RT 1:220 (customer VPN)
- RT 1:1000 (overlapping VPN)
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The figure shows how to implement overlapping VPNs between two customer central sites.
Only the configuration on the customer central site PE router has to be changed:
New virtual routing and forwarding (VRF) is created for the customer central site.
A new router distinguisher (RD) is configured for customer central site.
For the Customer A central site, the following occurs:
Import and export routes with RT 1:21 0 (customer routes)
Import and export routes with RT 1: 1000 (overlapping routes)
For the Customer B central site, the following occurs:
Import and export routes with RT 1:220 (customer routes)
Import and export routes with RT 1: 1000 (overlapping routes)

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-7

 -------.
Customer A (central) client can communicate with:
- All Customer A sites (customer VPN)
- Customer B central site (overlapping VPN)
Customer B (central) client can communicate with:
- All Customer B sites (customer VPN)
- Customer A central site (verlapping VPN)
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Because sites belonging to different VPNs do not share routing information, they cannot talk to
each other. The figure shows overlapping VPN data flow:
The simple VPN for Customer A contains routes that originate from the following:
A-Central site
A remote sites
The simple VPN for Customer B contains routes that originate from the following:
B-Central site
B remote sites
The overlapping VPN contains routes that originate from the following:
A-Central site
B-Central site
All Customer A sites can communicate with each other.
All Customer B sites can communicate with each other.
A-Central and B-Central can communicate with each other.
The customer A remote site cannot communicate with the customer B remote sites.
The customer A central site cannot communicate with the customer B remote sites.

Note If a site participating in more than one VPN is propagating a default route to other sites, it
can attract traffic from those sites and start acting as a transit site between VPNs, enabling
sites that were not supposed to communicate to establish two-way communication.

3-8 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Configure a new VRF instance for the central site:
- Import and export RTs for remote sites.
- Import and export RTs for overlapping sites .
Update BGP configuration:
- Set RD for the central site.
- Under the proper address family (IPv4 or IPv6), configure route redistribution.

To configure overlapping VPNs, the administrator has to do the following:

Configure a new VRF instance for the central site
Import and export route targets (RTs) for remote sites
Import and export RTs for the overlapping site
Update BGP configuration
Set RDs for the central site
Under the proper address family (IPv4 or IPv6), configure route redistribution

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-9

 vrf CustomerA-Cent vrf CustomerB-Cent
description Customer A Cent description Customer B Cent
address-family ipv4 unicast address-family ipv4 unicast
import route-target import route-target
1,210 1:220
1,1000 1:1000
export route-target export route-target
1,210 1:220
1,1000 1:1000

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The figure shows a Cisco lOS XR configuration example for configuring VRF instances.
Provider edge (PE) configuration is as follows:
vrf CustomerA-Cent
description Customer A Cent
address-family ipv4 unicast
import route-target
1:210
1:1000
export route-target
1:210
1:1000

vrf CustomerB-Cent
description Customer B Cent
address-family ipv4 unicast
import route-target
1:220
1:1000
export route-target
1:220
1:1000

vrf CustomerA
description Customer A
address-family ipv4 unicast
import route-target
1:210
export route-target
1:210

vrf CustomerB
description Customer B
address-family ipv4 unicast
import route-target
1:220
export route-target
1:220

3-10 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 router bgp 64500 router bgp 64500
vrf CustomerA vrf CustomerA
rd 1:210 rd 1:210
address-family ipv4 unicast address-family ipv4 unicast
redistribute connected redistribute connected

vrf CustomerB vrf CustomerB

rd 1:220 rd 1:220
address-family ipv4 unicast address-family ipv4 unicast
redistribute connected redistribute connected
vrf CustomerA-Cent vrf CustomerB-Cent
rd 1:211 rd 1:221
address-family ipv4 unicast address-family ipv4 unicast
redistribute connected redistribute connected

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The figure shows a Cisco lOS XR configuration example for configuring the Border Gateway
Protocol (BGP) process.
PE configuration is as follows:
router bgp 64500
vrf CustomerA
rd 1:210
address-family ipv4 unicast
redistribute connected

vrf CustomerB
rd 1:220
address-family ipv4 unicast
redistribute connected

vrf CustomerA-Cent
rd 1:211
address-family ipv4 unicast
redistribute connected

vrf CustomerB-Cent
rd 1:221
address-family ipv4 unicast
redistribute connected

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-11

Central Service VPNs
This topic describes central service VPNs and advanced VRF features.

Multiple VPNs need to share a

common set of servers:
VPN D - VPNs are called clients.
(Client)
Servers reside in central services
VPN E
VPN:
(Client)

Iv' - VPNs are called servers.

Clients from other VPNs cannot
communicate with each other.
Central SelVices
VPN
(SelVer'" ~~
6 ~

A ~
VPNC
(Client)

VPNA
(Client)
t VPN B
(Client)

C2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

A central service VPN is a topology with these characteristics:

Some sites ("server sites") can communicate with all other sites.
All the other sites ("client sites") can communicate only with the server sites.

This topology can be used in these situations:

The service provider offers services to all customers by allowing them access to a common
VPN.
Two (or more) companies want to exchange information by sharing a common set of
servers.
A security-conscious company separates its departments and allows them access only to
common servers.

3-12 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Import ( ; Client VPN routes:
Export
RT 1:220 ~ 4f!J Exported to the server site
VPN B
(Client) Export Server VPN routes:
RD 1:220 \ T1:501
Import Exported to client sites
Export
RT 1:21 0 Import ~ Exported to servers sites
~ RT1:502
No route exchange between
~~n~) ~ ~~.
R~xf:~~2
RD 1:210 RT 1:502
) Import
RT 1:501
client sites
Export

~
RT1:502

.
Central Services VPN
Export (Server)
RT 1:501 RD 1:500
,
~
Import
RT 1:501

Import
Export
RT 1:500

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

There is a specific routing model used to implement a central services VPN.

The figure illustrates the MPLS VPN routing model that is used to implement a central services
VPN:
Client site VPN routes are exported to the server site VPN.
Server site VPN routes are exported to client site VPNs
Server site VPN routes are also exchanged between other server sites.
There should be no route exchange and connectivity between client sites.

The figure shows a central services VPN (RD 1:500) with a set of common services. On client
site VPNs, routes are exported with RT 1:501 and imported with RT 1:502.
On the server site VPN, routes are exported with RT 1:502 and imported with RT 1:501.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-13

 Clients can talk to servers:
Client VRF contains server routes.
4!J
VPN B
4!J
(Client)
Servers can talk to clients:
RD 1:220
Server VRF contains client routes.

\
Clients cannot communicate:
Client VRFs do not contain routes
VPNA from other clients;
(Client) _ /
RD1:210~ Make sure that there is no
c1ient-to-client leakage across
Central Services VPN
(Server) server sites.
RD 1:500

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

In the central services VPN topology, the client VRF contains only routes from the client site
and routes from the server sites. This setup precludes the client sites from communicating with
other client sites.
A server VRF in this topology contains routes from the site or sites attached to the VRF and
also routes from all other client and server sites. Hosts in server sites can therefore
communicate with hosts in all other sites.

Note If the central site is propagating a default route to other sites, it can result in client sites
seeing each other through the CE router in the central site.

3-14 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Client sites:
- Use a separate VRF per client site.
- Usea unique RD on each clientsite.
- Import and export routes within customer sites.
- Export routes to server sites.
- Import routes from server sites .
Serversites:
- Use one VRF for each service type.
- Usea unique RD on each service type.
- Import and export routes within server sites.
- Export server site routes to clients.
- Import routes from client sites.

To configure a central services VPN, you need to address these requirements:

Client sites
Use a separate VRF per client site
Use a unique RD on each client site
Import and export routes within customer sites
Export routes to server sites
Import routes from server sites
Server sites
Use one VRF for each service type
Use a unique RD on each service type
Import and export routes within the server site
Export server site routes to clients
Import routes from client sites

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-15

 (;~ (C~
~VPNA~
(Client) Central Services VPN
RD 1:210
MPLS fJ------::-- (Server)
RD 1:500
~ (@ PE-CS-1
VPN B
(Client)
RD 1:220

vrf Cus tomerA vrf Server

address-family ipv4 unicast address-family ipv4 unicast
import route-target import route-target
1,210 1,500
1,502 1,501
export route-target export route-target
1,210 1,500
1,501 1,502

vrf CustomerS
address-family ipv4 unicast
import route-target
1,220
1,502
export route-target
1,220
1,501

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

The configuration example in the figure shows how to configure a central services VPN in a
Cisco IP NGN service provider network.
On PEl, two VRF instances are configured for Customer A and Customer B. On the PE-CS-I
router, the VRF server is configured.
Router PE-CS-I exports routes with RT I :502 and imports client routes with RT I: 50 1.
Customer VRF exports routes with RT I :50 I and imports server routes with RT 1:502.

3-16 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customers run a simple VPN.
Only A-Central and B-Central need access to central servers.
Solution:
- Combine a si mple VPN and central services VPN.
- Configure a separate VPN per customer.
- Configure a separate VRF for central servers.
- Configure a separate VRF for clients that need access to central servers (per
site).

In this design, some of the customer sites need access to the central server. All other sites just
need optimal intra-VPN access. The design is consequently a mixture of simple VPN topology
and central services VPN topology.
When integrating a central services VPN with a simple VPN, you need one VRF per VPN for
sites that have access to other sites in the customer VPN but that have no access to the central
services VPN. You need one VRF per VPN for sites that have access to the central services
VPN. Finally, you need one VRF for the central services VPN.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-17

 Combination of rules from:
- Overlapping VPN
- Central services VP N
Only central sites need access to central servers.
Configuration steps:
- Configure the customer VPN import-export RT in all VRFs participating in the
customerVPN.
- Configure a unique import-export RT in every VRF that is only a client of
central servers.
- Configure the central services import and export RTs in VRFs that participate
in the central services VPN.

To integrate central services and overlapping VPNs, you have to combine rules from
overlapping VPNs and central services VPNs.
The configuration steps are as follows:
Configure the customer VPN import-export RT in all VRFs that are participating in the
customer VPN.
Configure a unique import-export RT in every VRF that is only a client of central servers.
Configure the central services import and export RTs in VRFs that participate in the central
services VPN.

3-18 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Selective import:
- This feature allows you to specify additional criteria for importing routes into
theVRF.
Selective export:
- This feature allows you to specify additional RTs that are attached to exported
routes.

These advanced VRF features allow you to deploy advanced MPLS VPN topologies or increase
the stability of the MPLS VPN backbone:
The selective import feature allows you to select routes to be imported into a VRF based on
criteria other than the RT of the VRF.
The selective export feature allows you to attach specific RTs to a subset of routes that are
exported from a VRF. By default, the same RTs get attached to all exported routes.

Note The VRF route limit is also an advanced VRF feature on some platforms that allows you to
limit the number of routes that the customer-or other PE routers-can insert in the VRF.
This feature prevents undesirable consequences such as configuration errors or denial-of-
service (DoS) attacks.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-19

 VRF import criteria are more specific than just the match in RT:
- Import only routes with specific BGP attributes
- Import routes with specific prefixes or subnet masks
Route policy is used to make the route import selection more specific.

.1--
Use the import route-policy <name> command in VRF configuration
submode.

PE-l#
vrf Cus tomerA
address-family ipv4 unicast
import route-policy CustA-Policy
import route-target
1,210

export route-target
1,210

route-policy CastA-Policy
if destination in (192.168.1.0/24) then
pass
endif
end-policy
PE-2
C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

Selective route import into a VRF allows you to narrow the route import criteria. Selective
route import uses a route policy that can filter the routes selected by the RT import filter. The
routes imported into a VRF are Border Gateway Protocol (BGP) routes, so you can use match
conditions in a route policy to match any BGP attribute of a route. These attributes include
communities, local preference, multi-exit discriminator (MED), autonomous system (AS) path,
and so on.
The import route-policy command is combined with the RT import filter. A route must pass
the RT import filter first and then the import route policy. The necessary conditions for a route
to be imported into a VRF are as follows:
At least one of the RTs attached to the route matches one of the import RTs configured in
the VRF.
The route is permitted by the import route policy.

The figure shows an example in which an import route policy is used to match the IPv4 portion
of incoming VPNv4 routes and to import into the VRF only routes matching a certain prefix.
A configuration similar to this one could be used to accomplish the following:
Deploy advanced MPLS VPN topologies (for example, a managed router services
topology)
Increase the security of an extranet VPN by allowing only predefined subnetworks to be
inserted into a VRF, thus preventing an extranet site from inserting unapproved
subnetworks into the extranet

Note A similar function is usually not needed in an intranet scenario because all customer routers
in an intranet are usually under common administration.

3-20 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Routes from a VRF might have to be exported with different RTs:
- Export management routes with particular RTs .
An export route policy is used to set extended community RTs.

PE-l#
vrf Cus tomerA
address-family ipv4 unicast
import route-target
1,210

export route-policy ExportPol

export route-target
1,210

route-policy ExportPol
if destination in (192.168.1.0/24) then
set extcommuni ty rt 1: 555 addi ti ve
else
pass
endif
end-policy

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

Some advanced MPLS VPN topologies are easiest to implement if you can attach various RTs
to routes exported from the same VRF. This capability allows only a subset of the routes
exported from a VRF to be imported into another VRF. Most services in which customer
routers need to connect to a common server (for example, network management stations, voice
gateways, and application servers) fall into this category.
The export route-policy command provides exactly this functionality. A route policy can be
specified for each VRF to attach additional RTs to routes exported from that VRF. The export
route-policy command performs only the attachment ofRTs. It does not perform any filtering
functions.
Attributes attached to a route with an export route policy are combined with the export RT
attributes. If you specify export RTs in a VRF and set RTs with an export route policy, all
specified RTs will be attached to the exported route.

Note The export route policy provides functionality that is almost identical to that of the import
route map, but applied to a different VRF. Any requirement that can be implemented with an
export route policy can also be implemented with an import route policy. However, the
implementation of export maps can be more complicated and difficult to manage.

In the figure, the configuration is implemented with an export route policy.

Note Depending on when you configure the export map command, you might need to use the
clear bgp command to force the existing BGP session to propagate the extended
communities.

In this example, routes from a certain address block are marked with an additional RT in the
originating VRF and are automatically inserted into the receiving VRF based on their RT.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-21

Managed CE Router Service
This topic describes managed CE router service.

Service providers use network management VP N to manage the CE

routers of all VPNs:
- Central server NMS needs access to the loop back address of all CE routers.
- S imilarto central services and simple VRFs
- CE routers participate in the central services VPN.
- Only loopbackaddresses of the CE routers are exported into the central
services VPN.

If the service provider is managing the customer routers, it is convenient to have a central point
that has access to all CE routers but does not have access to the other destinations at the
customer sites. This requirement is usually implemented by deploying a separate VPN for
management purposes. This VPN needs to see all the loopback interfaces of all the CE routers.
All CE routers need to see the network management VPN. The design is similar to that ofthe
central services VPN; the only difference is that you mark only loopback addresses to be
imported into the network management VPN.
Managed CE router service features include the following:
The central server network management system (NMS) needs access to the loopback
address of all CE routers
It is similar to central services and simple VRFs
CE routers participate in the central cervices VPN
Only the loopback addresses of the CE routers are exported into the central services VPN

3-22 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Create one VRF per customer VPN per PE router.
- Assign the same RD to each customerVRF.
Create an NMS VRF on the central services PE router:
- Assign a unique RD to the NMS VRF.

Customer A Customer B
RD 1:210 NMS Server RD 1:220
RD 1 :500

The configuration overview is as follows:

Create one VRF per customer VPN per PE router
Assign the same RD to each customer VRF
Create an NMS VRF on the central services PE router
Assign a unique RD to the NMS VRF

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-23

 vrf Cus tomerA
address-family ipv4 unicast
import route-target vrf NM'S Servers
1,210 address-family ipv4 unicast
1,500 import route-target
export route-policy MGMT_Pol 1,500
export route-target 1,501
1,210 export route-target
1,500
route-policy MGMT_Pol
if destination in (192.168.1.0/24) then
set extcommunity rt 1:501 additive
else
pass
endif
end-policy

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

You can have a configuration for a customer VRF with differentiated RT export for loopback
addresses.

The figure shows this example. An export route policy is used to match one part of the IP
address space and attach an additional RT to the routes within this address space (CE router
loopback addresses).

Note The routing protocol between PE and CE routers must be secured (with distribute lists or
prefix lists) to prevent customers from announcing routes in the address space dedicated to
network management; otherwise, customers can gain two-way connectivity to the network
management station.

The CE router loopback addresses are then imported into the server VPN based on the
additional RT attached to them during the export process.

Note This design allows client sites to send packets to the network management VPN regardless
of the source address. Special precautions should be taken to protect the network
management VPN from potential threats and DoS attacks coming from customer sites.

3-24 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary points that were discussed in this lesson.

Overlapping VPNs are used to provide connectivity between segments

in twoVPNs.
Central services VPNs offer the following:
- Customers can access common services.
- Customers cannot communicatewith each other.
- Route pol icies can be used for selective route im port and export.
Service providers can access the management loopback interface ofCE
routers. Service providers use:
- NMS VRF
- Export route policy

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-25

3-26 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 21

Implementing Internet Access

and MPLS Layer 3 VPNs
Overview
Integrating Internet access with a Multiprotocol Label Switching (MPLS) VPN solution is one
of the most common service provider business requirements. This lesson provides a good
understanding of underlying design issues, several potential design scenarios, and some sample
configurations. Various topologies and implementation methods are discussed, along with ways
to separate Internet access from VPN services.

Objectives
Upon completing this lesson, you will be able to describe common customer Internet
connectivity scenarios and identify design models for combining Internet access with MPLS
Layer 3 VPN services. You will be able to meet these objectives:
Describe common customer Internet connectivity scenarios and identify design models for
combining Internet access with MPLS Layer 3 VPN services
Describe implementation of the Internet access service totally separate from MPLS Layer 3
VPN services
Describe implementation of the Internet access solutions in which Internet access is
provided as a separate VPN
Internet Access Models with MPLS VPNs
This topic describes common customer Internet connectivity scenarios and identifies design
models for combining Internet access with MPLS Layer 3 VPN services.

Internet routing is usually performed via the BGP table of the MPLS VPN
network of the service provider.
By default, the VRF sites:
- Can communicate only with devices in otherVRF sites of the same VPN
- Cannot communicate with devices in the global routing space
There is potential security risk in providing Internet connectivity:
- Firewalls are used to ensure the highest possible level of security.

Internet routing is usually performed via the Border Gateway Protocol (BGP) table of the
MPLS VPN network of the service provider. This BGP table is in the global routing space, not
in the virtual routing and forwarding (VRF) context. By default, the VRF sites can
communicate only with other VRF sites in the same VPN, not with anything in the global
routing space. Therefore, something must be done to provide Internet access (global context) to
the customer edge (CE) routers (VRF context). The following subtopics explain how to provide
Internet access to VRF sites. Internet access is possible only for those customer IP subnets that
are not from the private IP addressing space (RFC 1918).

Note As soon as the VPN has Internet connectivity, a potential security risk exists. Take the
proper steps-such as filtering and using a firewall-to ensure the highest possible level of
security.

3-28 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customer connects to the Internet through a central site firewall:
- Deals with security issues
- Provides NAT or proxy services as needed
Internet traffic goes across the central site:
- Traffic flow is not optimal.

Customer A (1) Customer A (2)...)

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Classical Internet access is implemented through a central firewall that connects the customer
network to the Internet in a secure fashion.
The customer network and the Internet are connected only through the firewall. The addressing
requirements for this type of connection are simple. The customer is assigned a small block of
public address space that is used by the firewall. The customer typically uses private addresses
inside the customer network if they are using IPv4 addressing. The firewall performs Network
Address Translation (NAT) between the private addresses of the customer and the public
addresses that are assigned to the customer by the ISP. Alternatively, the firewall might
perform an application-level proxy function that also isolates private and public IP addresses.
If a customer is using IPv6 addressing, there is no need for NAT, but the customer still needs
other functions that a firewall provides.
Several benefits are associated with this design. The setup is well known, and the expertise
needed to implement it is simple and straightforward. Only one interconnection point between
the secure customer network and the Internet needs to be managed.
The major drawback ofthis design is the traffic flow. All traffic from the customer network to
the Internet passes through the central firewall. Although this flow might not be a drawback for
smaller customers, it can be a severe limitation for large organizations with many users,
especially when they are geographically separated.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-29

 o Customers have Internet access directly from every site.
o Optimum traffic flow for Internet traffic
o Each site has to deal with security issues:
- Managed firewall offered by service provider
- Customer firewall

.
Customer A (1)

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Some customers find the traffic flow limitations of the central firewall setup too limiting. To
bypass the limitations of Internet access through a central firewall, some customers use designs
in which each customer site has its own independent Internet access.
This design solves traffic flow issues, but the associated drawback is higher exposure. Each site
needs to be individually secured against unauthorized Internet access, leading to the increased
complexity of managing a firewall at every customer site.
To achieve Internet access from every customer site, each CE router must forward VPN traffic
toward other customer sites and forward Internet traffic toward Internet destinations. The two
traffic types are usually sent over the same physical link, but over different logical links, to
minimize costs.
For customers that do not want the complexity of managing their own firewalls, a managed
firewall service offered by the service provider can help address the security issues of Internet
connectivity.

3-30 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Customer chooses an ISP and selects services.
User can access different services offered by different service providers.
Internet access backbone:
- Provided by NSP
- Used to interconnect customerwith service provider

Network service
Provider

Backbone

~ SelVice Provder Z

A service provider might provide wholesale Internet access from a range of upstream ISPs to
satisfy the connectivity and reliability requirements of various customers.
The selection of upstream ISPs and the corresponding configuration processes should therefore
be as easy as possible for the service provider. From an Internet perspective, customers A, B,
and C are connected to ISP X or ISP Y. The IP address space that the customer uses should be
allocated from the block of addresses that is administered by the selected ISP. The service
provider that provides wholesale Internet access might need to use a different address for each
upstream ISP.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-31

 Internet used to be the most popularservice.
Clients expect different services from their service providers now:
- Internet, video, IP telephony, cloud services, and so on
Cisco IP NGN architecture supports multiple services in a common
backbone.

.
Customer A
(CentraO
VPN, Internet
.
Customer A (1]-J
VPN, Internet
/
Service Provider X
Internet, IP
81ephony, Video
J

Network service Service Provider Y

Cu.erB Provider -- Internet, Ooud,
IP Telephony, IP Telephon

~
I;t /
SelVeeProvoer Z
CustomerC ...) VPN, Internet, j
Interne( Cloud IP TelephonYI

Because Internet access is one of the most popular services that service providers offer their
customers, many service providers offer Internet access as well as MPLS VPN service on their
shared backbone. Integrating Internet access with an MPLS VPN solution is one of the most
common service provider business requirements. A background of common customer Internet
connectivity scenarios will help in assessing possible implementations.

3-32 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Two major design models:
- Internet access through global routing
- Internet access as a separate VPN service
Internet access through route leaking is not an appropriate model for
service providers:
- Scalability problems

Network designers who want to offer Internet access and MPLS VPN services on the same
backbone can choose between these two major design models:
Internet access that is implemented through global routing on the provider edge (PE)
routers and that is not a VPN service
Internet access that is implemented as a separate VPN in the ISP network
In both cases, security should be the most important concern for customers when they connect
to public networks. Customers should isolate private VPNs from Internet traffic, either
physically (on a separate interface) or on a subinterface. Appropriate firewall support, either in
a dedicated device or integrated in the router Cisco lOS Software, is a necessity. Depending on
the network addressing, NAT will be needed for most customers.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-33

 Separate interface for VPN a nd Internet
- In global routing table
- Static default routing on aPE
- B GP between CE and PE
Benefits:
- Well-known setup (equ ivalent to classical Internet service)
- Easy to implement
- Offers a wide range of design options
Drawback:
- Requires separate physical links or WAN encapsulation that supports
su bi nterfa ces

Implementing Internet access through global routing is identical to building a traditional IP

backbone that offers Internet services. Depending on whether the customer sites need full
Internet routes, static routes or BGP is used for routing to the Internet. For full Internet routes,
BGP is deployed between the PE routers and the PE Internet gateway to exchange Internet
routes, and the global routing table on the PE routers is used to forward the customer traffic
toward Internet destinations. The PE routers might or might not have the full Internet routing
table. The PEs and the provider Internet gateway are in the same Internal BGP (IBGP) area.
The PE routers also use Multiprotocol BGP (MP-BGP) to support VPNs for their customers.
The customers reach the global routing table by using a separate logical link for Internet access.
Internet access through global routing on separate logical links is easy to set up; it is the
equivalent ofthe classical combination ofInternet and VPN services that many customers use
today. This setup is also compatible with all the Internet services required by some customers
(for example, the requirement to receive full Internet routing from a service provider).
The drawback of this design is the increased complexity, or cost, of the PE-CE connectivity.
Separation ofInternet and VPN connectivity requires either two separate physical links or a
single physical link with encapsulation that supports subinterfaces; for example, Frame Relay,
PPP or using Ethernet links and 802.1 Q encapsulation.

3-34 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Implementation through a separate VPN
Benefits:
- Provider backbone is isolated from the Internet.
- Increased security
Drawbacks:
- All Internet routes are carried as VPN routes.
- Scalability problems-full Internet routing table in VPN

For a service provider, implementing Internet access through a separate VPN is similar to
offering another managed VPN service.
The major benefit of implementing Internet access as a separate VPN is increased isolation
between the provider backbone and the Internet-which results in increased security for the
provider. The flexibility ofMPLS VPN topologies also provides for some innovative design
options that allow service providers to offer services that were simply impossible to implement
with pure IP routing.
The obvious drawback of running the Internet as a VPN in the MPLS VPN architecture is the
scalability of such a solution. An Internet VPN cannot carry full Internet routing because of the
scalability problems that are associated with having all the Internet routes inside a single VPN.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-35

 Not a recommended design:
- Formerly used in corporate environments
- Internet access across corporate VPN:
Leaking routes between VRF and global routing table
Benefits:
- Does not use a separate connection for Internet traffic
Drawbacks:
- Insecu re---Internet traffic rnixed with VPN traffic
- Hard to apply security policies
- Scalability problems-hard to irnplernentfullinternet routing

Another variant for Internet support with MPLS VPNs is to provision route leaking into the
global routing context. Although not a recommended design, this option is briefly discussed to
show an alternate practice that has been used in the industry.
Some customers might want to obtain Internet access across their corporate VPN by leaking
routes between the VRF and global routing tables.

Caution For security reasons, this approach is not recommended. Bringing in Internet traffic by using
the corporate VPN is not a good practice and negates the isolation of the corporate VPN.

With route leaking, the customer site uses a static default route in the VRF table that points to
the global next-hop address of an Internet gateway. Any packets that use the default route leave
the VPN space and are routed according to the global routing table at the PE that is the next-
hop router. This feature allows leaking of VPN packets into the global address space.

Note This approach is not recommended and is not discussed further in this course. This option is
discussed briefly only to show an alternate practice that has been used in the industry.

3-36 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Separate Internet Access and VPN Services
This topic describes implementation of the Internet access service totally separate from MPLS
Layer 3 VPN services.

Customer A (3)

~ VPN

I::.
IntemetGW
Shared

'~~
PE1 Backbone

MPLSVPN&

E3

V!

Customer A
(Central)
\7R & Internet

1C12012Ciscoandioritsatrlllllte!l.A1I~ghts",served.

The classical Internet access design for a customer is based on a separate Internet access model.
One central customer site has connectivity to the Internet and provides access to the rest of the
customer sites. The central site either connects through a firewall or runs the Cisco lOS
Firewall Feature Set.
In the shared service provider backbone, the PE routers have full or partial Internet routes to
offer the customer. The Internet gateway of the provider is in the same IBGP domain as the
provider and PE routers.
This design model can easily map to a customer with an MPLS VPN implementation. In this
example, the customer network has been interconnected with an MPLS VPN. A central CE
router has Internet connectivity and provides Internet access, through a firewall, for all the sites
in the customer network.
This traditional Internet access implementation model provides maximum design flexibility
because the Internet access is completely separated from the MPLS VPN services. However,
the limitations of traditional IP routing prevent this implementation method from being used for
innovative Internet access solutions, such as wholesale Internet access.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-37

 o Separating physical links forVPN and Internet is sometimes
unacceptable because of high cost.
o Sub interfaces can be used:
- Over WAN links
Frame Relay
ATM
- Over LAN links (802.1 Q)
o A tunnel interface could be used over a VRF-aware tunnel, so thatVPN
traffic does not run over a global tunnel.

Instead of separate physical links for VPN and Internet traffic, subinterfaces can be used to
create two logical links over a single physical link. Subinterfaces can be configured only on
WAN links that use Frame Relay or ATM encapsulation (including xDSL) and on LAN links
that use any VLAN encapsulation (802.1 Q).
For other encapsulation types, a tunnel interface can be used between the CE router and the PE
router. Depending on the router platform and Cisco lOS Software version, virtual routing and
forwarding (VRF)-aware tunnels are now supported.
VRF-aware tunnels remove the need for the endpoints of the tunnel to be in a global
routing table.
Without a VRF-aware tunnel, MPLS VPN traffic would need to be tunneled across the
Internet interface.

Note Further information on VRF-aware tunnels is outside the scope of this course.

3-38 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 vrf CustomerA
Customer A (1) ~ address-family ipv4 unicast
VPN import route- target
10.10.1.0/24 1;210
export route-target
1;210

PE1
..... I
interface GigabitEthernetO/l
no ip address
I
interface Gigabi tEthernetO/l. 2
description Internet
encapsulation dot1Q 2 native
ip address 172.16.10.1 255.255.255.252
I
interface Gigabi tEthernetO/l. 3
description MPL VPN
ip vrf forwarding CustomerA
interface G:Lgab:L tEthernetO/l. 2 encapsulation dot1Q 3 native
descr:Lption Internet ip address 192.168.16.1 255.255.255.252
encapsulat:Lon dot1Q 2 nat:Lve I
ip address 172.16.10.2 255.255.255.252 router static
I address-family ipv4 unicast
interface Gigabi tEthernetO/l. 3 209.165.201.0/27 172.16.10.2
description MPLS VPN I
encapsulation dot1Q 3 router bgp 64500
ip address 192.168.16.2 255.255.255.252 address-family ipv4 unicast
I redistribute static
ip route 0.0.0.0 0.0.0.0 172.16.10.1
ip route 10.10.0.0 255.255.0.0 192.168.16.1

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Using static routing on the CE and PE routers is the simplest and most common implementation
for providing Internet access. The figure illustrates a configuration that is used to implement
Internet access through two 802.1 Q subinterfaces with static routes.
In this simple example, the Customer A router does not need to receive full Internet routing. To
reach the Internet, the Customer A router just needs a default static route to PE3. PE3 has a
route to the Internet through the Internet gateway, as well as a static route for the customer A
subnets that point to the Customer A router. The full Internet routing table needs to be present
only on the Internet gateway.
The following configuration steps are performed:
1. The customer VRF instance is created for private MPLS VPN.

2. The VPN subinterface is created and associated with the proper VLAN. The subinterface is
added to the customer VRF.

3. Another subinterface is created for Internet access.

4. On the PE router, configure static routes for customer public address space pointing to
customer next hop. Redistribute these routes in BGP routing protocol.

In addition to the PE configuration, the CE router implements a default static route that points
to the PE router.

Note On the CE-Central router, distribution of the default route might be needed so that remote
sites can also access the Internet. Issues of security and private addresses would need to
be resolved.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-39

 vrf CustomerA
~ address-family ipv4 unicast
VPN
Customer A (1)
10.10.1.0/24-1
import route- target
1;210
export route-target
Intemet GW' , IBGP 1;210
Shared \ I
PE1 ............... Backbone 1 interface Gigabi tEthernetO/l. 2
...... I description Internet
Customer A (2) encapsulation dot1Q 2
VPN __ ~M:~~VPN_:" / I
ip address 172.16.10.1 255.255.255.252
10.J,O.:1Jl!24
interface Gigabi tEthernetO/l. 3

i~7"
encapsulation dot1Q 3
interface Gigabi tEthernetO/l. 2 vrf Customer-A
description Internet ip address 192.168.16.1 255.255.255.252
encapsulation dot1Q 2 I
ip address 172.16.10.2 255.255.255.252 router bgp 64500
I address-family ipv4 unicast
interface Gigabi tEthernetO/l. 3 neighbor 172.16.10.2
description MPLS VPN remote-as 64503
encapsulation dot1Q 3 update-source Gigabi tEthernetO/O/O/O. 2
ip address 192.168.16.2 255.255.255.252 address-family ipv4 unicast
I route-policy pass in
router bgp 64503 route-policy Only_ Defaul tout
network 209.165.201.0 mask 255.255.255.0 default-originate
neighbor 172.16.10.1 remote-as 64500 next-hap-self
I I
ip route 209.165.201.0 255.255.255.0 null route-policy Only_ nefaul t
I if destination in (0.0.0.0/0) then
pass
endif
end-policy

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

In this example, Customer A is using a dynamic routing protocol to establish Internet

connectivity.
BGP routing protocol is configured on the Customer A router. Customer A advertises its own
public address space to External BGP (EBGP) neighbor. It is important that customer public
address space is inserted into the customer routing table. If it is not in the routing table, the
network is not advertised using BGP.
On the PE router, the service provider has to configure the VRF instance and subinterface for
the private MPLS VPN of the customer and another subinterface for Internet access.
BGP is used for route exchange between the customer and service provider. The service
provider can originate the default route to the customer peer. The service provider can also
filter advertised routes and send only the default route to the customer. A route policy is used to
filter routes on Cisco lOS XR routers, and prefix lists are used to filter advertised routes on lOS
routers.

3-40 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Every CE router needs two links (or subinterfaces).
Complex network setup

..
Expensive solution

Customer A (1)
Customer A (3)
VPN
V. N & Internet / / VPN
~VPN& Internet
IntemetGW
Shared

'~~
PE1 Backbone

'"~""A[" c~
VPN
'VRN & Interne,tJ
_ Customer A (4)
VPN
VPN & Internet

Customer A

VP~~~~~;n.t'
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Another option is to provide separate Internet access at every customer site. In this case, two
physical (or logical) links between every CE router and its PE router would be needed. This
design often becomes too complex or too expensive to implement. Issues such as customer
route propagation to the Internet and securing access at multiple access points would need to be
resolved.

Note The allowas-in feature might need to be configured on the PE router if the customer is
propagating individual site routes to the Internet through BGP.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-41

 Benefits of separate Internet access:
- Well-known model
- Supports all customer requirements
- Allows all Internet service implementations
Drawbacks of separate Internet access:
- Requires separate physical link
- PErouters must be able to perform Internet routing
Potentially carry full Internet routing table
Wholesale Internet access cannot be implemented in this model.

The benefits of a separate Internet access design model are as follows:

The model is well known and widely understood.
The model supports all customer requirements, including multihomed customer
connectivity with full Internet routing.
The model allows all Internet service implementations, including BGP sessions with
customers.

The drawbacks of this model are as follows:

The model requires two dedicated physical links between the PE and the CE router or
specific WAN or LAN encapsulations that might not be suitable for all customers.
The PE routers must be able to perform hop-by-hop Internet routing and either use the
default route to reach the Internet or carry the full Internet routing table.
Advanced Internet access services (centrally managed firewall service or wholesale Internet
access service) cannot be realized with this model.

3-42 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Internet Access as a Separate VPN
This topic describes implementation of the Internet access solutions in which the Internet
access is provided as a separate VPN.

Service provider gateway is connected as a CE routerto the MPLS VPN

backbone.
Global Internet routing table is very big:
- Only default route and some specific regional routes are distributed to the
MPLS VPN network.
Many service providers on same network backbone:
- Customer can chose service provider.
- Customer site is assigned to VRF of service provider.

The MPLS VPN architecture can provision a separate VPN to provide Internet access for VPN
customers. The service provider defines the Internet VPN and can use different MPLS VPN
topologies to implement various types ofInternet access. Under this design model, the provider
Internet gateways appear as CE routers to the MPLS VPN backbone. Customer Internet access
is enabled by using a dual VPN topology that supports both an Internet VPN and a customer
VPN across separate customer interfaces.
In this design, the Internet VPN should not contain the full set of global Internet routes because
that would make the solution completely nonscalable. The provider Internet gateway routers
should announce a default route toward the PE routers. To optimize local routing, the local and
regional Internet routes should be inserted in the Internet VPN.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-43

 --_._~.
Internet gateway has full Internet routing table:
- Only subset of all routes sent to customers
Internet gateway acts as a CE router.
Internet VPN is used for Internet access.
Customers are assigned to Internet VPN.
Internet

"'lnternet GW

liJ ~
CustomerS (1) ~

~N
P E-GW
Shared
PE1 Backbone
PE4

CustomerS
(Center)
~2
~;
- . - . -.......-. " " , . - - -
Customer A
VPN, Internet ~-----liiiIiJ. (Center)
VPN, Inta.wet

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

When the service provider implements Internet access as a separate VPN, the Internet backbone
is carried on a VPN, which is isolated from the provider backbone. This topology results in
increased security for the provider backbone because Internet hosts can reach only PE routers,
not the core provider routers. The VPN customers are connected to the Internet simply through
an additional VRF instance at the PE.
Internet gateway acts as a CE router and holds the full BGP routing table. Only the subset of
routes and the default route are advertised to clients assigned to Internet VPN.

3-44 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 PE-GW: InternetGW:
vrf Internet interface GigabitBthernetO/1
description Internet ip address 172.16.255.2 255.255.255.252
I
address- family ipv4 unicast
import route-target router bgp 64510
1;2000 address - family ipv4 unicast
I
I
export route- target neighbor 172.16.255.1
1;2000 remote-as 64500
I
update- source Gigabi tBthernetO/1
address-family ipv4 unicast
interface Gigabi tBthernetO/1
route-policy pass in
vrf Internet
ip address 172.16.255.1 255.255.255.252 route-policy Only Default out
default originate-
I

~
I
router bgp 64500
route-policy Only Defau1 t
vrf Internet
rd 1;2000 if destination in (0.0.0.0/0) then
address - family ipv4 unicast pass

....
endif
I
end-policy
neighbor 172.16.255.2
remote-as 64510 '------ ------'
update- source Gigabi tBthernetO/1 1~2162551 \

~
172.162552
address - family ipv4 unicast
route-policy pass in
route-policy pass out
next-hop-se1f
I --
PE-GW .......... _
--~
BGP Intemet GW
Intemet

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The figure shows a sample configuration of the Internet gateway router with default route
advertisement.
The Internet gateway should be able to route traffic to the Internet. An EGBP session is
established with the PE gateway router. Only the default route is advertised to the PE gateway
router.
On the PE gateway router, a new VRF instance for Internet is used. The interface facing the
Internet gateway is assigned to the Internet VRF. In the BGP process, the Internet VRF has to
be enabled and appropriate address families have to be activated (such as IPv4 and IPv6). The
PE gateway router distributes the default Internet route among the other PE routers in the
MPLS VPN network.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-45

 Se
PEl:
vrf CustomerA Intemet
address-family ipv4 unicast
import route- target

'f'
1;210
export route-target
1;210
I
interface Gigabi tEthernetO/l. 2
description Internet
encapsulation dotlQ 2 PE-GW
ip address 172.16.10.1 255.255.255.252
I
router bgp 64500 ' - MPLS
address-family ipv4 unicast
I
address-family vpnv4 unicast
I

Q'"
I
vrf Internet
rd 1;2000
neighbor 172.16.10.2
remote-as 64503
address-family ipv4 unicast
Customer A
network 0.0.0.0/0
Internet, VPN

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The classical Internet access model can easily be implemented with the Internet VPN over the
MPLS VPN backbone, The link between a PE router and the Internet gateway router is
assigned to the Internet VRF, as discussed previously. The Internet gateway announces a
default route to the Internet.
One link between the PE router and each central customer router is assigned to the customer
VRF, and one is assigned to the Internet VRF.
In this example, PE 1 connects the Customer A router to both the Customer A VRF and the
Internet VRF.

3-46 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Allintemet gateways advertise
routes.
GW1 IBGP GWJ
Internet gateways are connected to
the same VRF. .~.
BGP metric is used to select the best
route to the Internet.
MED is used to define the primary
Internet gateway.
'''' ( ,,~
lIJ PE-GW2
~lIJ
PE-GW1 PE-GWJ

WLS

lIJPE3
PE2

Redundant Internet access is easy to achieve when the Internet service is implemented as a
VPN in the MPLS VPN backbone:
Multiple Internet gateways (acting as CE routers) need to be connected to the MPLS VPN
backbone to ensure router and link redundancy.
All Internet gateways advertise the default route to the PE routers, resulting in routing
redundancy.
The Internet gateways also announce local Internet routes. Because these routes are
announced with different BGP attributes-most notably multi-exit discriminator (MED)-
the PE routers select the proper Internet gateway router as the exit point toward those
destinations.
The MED attribute can also be used to indicate the preferred default route to the PE routers.
In this setup, one Internet gateway router acts as a primary Internet gateway, and the other
Internet gateway router acts as a backup.
The redundancy that has been established so far covers the path between customer sites and
the Internet gateway routers. A failure in the Internet backbone might break the Internet
connectivity for the customers if the Internet gateway routers announce the default route
unconditionally. Conditional advertisement of the default route is therefore configured on
the Internet gateway routers, which announce the default route to the PE routers only if the
Internet gateway routers can reach an upstream destination.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-47

 Internet VRF is configured on
every location.
Adds complexity Intemet
Customer A (3)

~:~
Firewall on every site: Internet, VPN

- Managed firewall can be used.

PE-GW

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Multisite customer Internet access can be implemented by configuring the Internet VRF at
every location. This solution adds complexity for the customer because firewall and Network
Address Translation (NAT) support might be needed at every site, unless the service provider
offers a central managed firewall service.

3-48 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 A separate VPN is created for each upstream ISP.
Each ISP gateway announces the default route to the VPN.
Customers are assigned into the right VRF:
- VRF assignment corresponds to ISP selection.
ISP change is easy for administrator:
- Only VRF has to be changed.

,.
Customer A
Service Provider X
Internet, IP ~
...... (Central) ~ elephony, Video

,.
~FW. Internet

,~c;st~':::~B_ ---
Network Service
Provider --
Service Provider Y
Internet, Cloud,
IP Telephony
~etl
~
/
Service Provider Z
CustomerC VPN, Internet,-,
1 ternet, Cloud IP Telep.hony

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Wholesale Internet access is implemented by creating a separate VPN for every upstream ISP.
Acting as a CE router toward the MPLS VPN-based Internet access backbone, the Internet
gateway of the upstream ISP announces a default route, which is used for routing inside the
VPN.
Customers are tied to upstream service providers simply by placing the PE-CE link into the
VRF that is associated with the upstream service provider. Changing an ISP becomes as easy as
reassigning the interface into a different VRF and managing address allocation issues.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-49

 Benefits:
- Supports all Internet access service types
- Easy to make changes
- Can support customer requirements
Drawbacks:
- Full Internet routing cannot be carried in the VPN:
Suboptimal routing
- Overlapping Internet and VPN backbone design requires special care.

Internet access that is implemented as a separate VPN has the following benefits:
This design model supports all Internet access services, ranging from traditional Internet
access to innovative services such as wholesale Internet access.
This design also supports all customer requirements, including full Internet routing on
customer routers through an EBGP multihop session with the Internet gateway.

Internet access that is implemented as a separate VPN has the following drawbacks:
Full Internet routing cannot be carried inside a VPN; therefore, default routing toward the
Internet gateways needs to be used, potentially resulting in suboptimal routing.
The Internet backbone gateway router is positioned as a CE router connected to the MPLS
VPN backbone. If the service provider runs Internet service and MPLS VPN service on the
same set of routers, the interconnection between the two services requires special
considerations.
The benefits of the separate VPN design far outweigh the limitations.

3-50 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary points that were discussed in this lesson.

Internet access types include the followi ng:

- Classical Internet access
- M ultisite Internet access
- Wholesale Internet access
Two recommended service provider designs are as follows:
- Global routing (global routing table is used for Internet routing)
- Internet service as a separate VPN
Wholesale Intemet access is easy to implement when you use Internet
service as a separate VPN.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-51

3-52 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 31

Introducing MPLS Interdomain

Solutions
Overview
Deployments of Multiprotocol Label Switching (MPLS) have become routine in large-scale
global networks, which demand solutions to complex business and network problems. There
are two primary components of the Cisco lOS MPLS Interdomain Solution: the
interautonomous system (inter-AS) and Carrier Supporting Carrier (CSC).

Objectives
Upon completing this lesson, you will be able to introduce the Cisco lOS MPLS Interdomain
solutions with different implementation options. You will be able to meet these objectives:
Describe MPLS interdomain solutions
Describe the CSC feature
Describe inter-AS MPLS models
MPLS Interdomain Solutions
This topic describes MPLS interdomain solutions.

Companies need MPLS service delivered all over the world.

Support forVPNs that cross AS boundaries
Two basic types of service provider design:
- esc
Hierarchical MPLS VPN design
Using other service providers for MPLS backbone
- Inter-AS
Peer-to-peer type model
Peering with neighboring service providers

Deployments of MPLS have become routine in large-scale global networks, which demand
solutions to complex business and network problems. There are two primary components of the
Cisco lOS MPLS Interdomain Solution: inter-AS and CSc.
Inter-AS is a peer-to-peer type model that allows the extension ofVPNs through multiple
provider or multidomain networks. This solution enables service providers to peer up with one
another and offer end-to-end VPN connectivity over extended geographical locations for those
subscribers who may be out of reach for a single provider.
CSC is a hierarchical VPN model that allows small service providers, or customer carriers, to
interconnect their IP or MPLS networks over an MPLS backbone. This eliminates the need for
customer carriers to build and maintain their own MPLS backbone.
Both inter-AS and CSC can construct scalable networks that help maintain network
segmentation based on internal organizational or operational boundaries.

3-54 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 IP Infrastructure Layer

o MPLS interdomain solutions are part of the Cisco IP NGN infrastructure

layer.
o IP edge devices run MPLS, BGP, or IGP.
o IP core devices run MPLS.
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

CSC is configured in the service provider core and edge network. It is part of the Cisco IP
Next-Generation Network (NGN) infrastructure layer.
IP core devices run MPLS and IP edge devices run MPLS, Border Gateway Protocol (BGP),
and some interior gateway protocol (IGP) routing protocols.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-55

 Hierarchical MPLS VP N:
- Backbone provider-first-Ievel service provider
- Customercarrier-second-Ievel service provider
esc provides MPLS VPN service to other service providers.
A large service provider acts as the backbone for smaller service
providers.
The customercarrier can be an ISP or MPLS VPN provider.

PE1
S PE2

Cus10mer Customer
Carrier Carrier
)
pop lite cse-CE1 CSC-CE2 pop lite

A carrier network carries traffic between customer sites. Large service providers can
interconnect carrier networks of smaller service providers. In this scenario, a smaller service
provider acts as a customer for a larger service provider.
esc provides MPLS VPN service to other service providers. esc creates a hierarchical
structure with a first-level service provider as the backbone carrier and a second-level service
provider as customer carrier. Many customer carrier sites, also called point of presence (POP)
sites, can be interconnected using the backbone carrier.

3-56 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Multiple customer carriers can be connected to a single CSC backbone.
Both VPN and Internet services can be provided.
Customercarriers do not have to operate their own long-distance
network.
Different addressing schemes can be used by different carriers.
Any link type supported by MPLS can be used.
There are no end-user routes in the CSC backbone.

esc has many benefits:

A single esc backbone can be used to connect many POP sites.
An MPLS VPN can be used to separate traffic from the POP sites of different carriers.
A single esc backbone can provide services to both VPN service providers and to the ISP.
The customer service providers do not have to operate their own long-distance network.
They purchase that service from the esc backbone carrier.
Different customer carriers can use different addressing schemes. The customer carriers
will be in separate MPLS VPNs inside the esc backbone.
Any link type that is supported by MPLS can be used inside the esc backbone and as
access links between the esc backbone and customer carriers.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-57

 Packets from POP1 to POP2 are propagated along a label-switched
path from CE1 to CE2.
PE and CSC-CE routers must exchange route or label information.
Backbone carrier does not carry routing information of end customers.

Customer Customer
Customer . .-----,t=:li:S Carrier Carrier
__._A_.._
CE1 CSC-PE p~.e-cE1

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The CSC architecture relies on the presence of an MPLS VPN. The CSC backbone is providing
an MPLS VPN service to which the customer carriers are connected as VPN sites. MPLS is
used between the CSC backbone provider edge (PE) routers and the VPN sites ofthe customer
earners.
Virtual routing and forwarding (VRF) tables are enabled on the CSC PE routers. The label
exchange between PEl and PE2 establishes a label-switched path (LSP) from CEI via the CSC
backbone to CE2. Another LSP is also established in the other direction.
The Customer carrier can now tunnel packets between POP I and POP2 using the LSPs. The
CSC backbone does not have to know about end-user sites and their IP addresses.

3-58 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 . esc backbone carrier must support MPLS VPNs.
. esc customer carrier can exchange labels:
- Using IGP and LDP:
o MPLS is enabled on link between backbone carrier and customer carrier.
o IGP is used for route exchange.
- Using MP-BGP:
o MP-BGP is used for label and route distribution.

To support ese, the ese backbone must support MPLS VPNs.

The customer carrier can connect to the backbone carrier in the following ways:
Using IGP and Label Distribution Protocol (LDP)
MPLS has to be enabled on the link between the backbone carrier and the customer
carrier. LDP is used for label distribution.
IGP is used for route exchange.
Using Multiprotocol BGP (MP-BGP)
MP-BGP is used for label and route exchange. There is no need for LDP.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-59

 ~-" liiii'"
Most MPLS VPN systems are deployed in one AS.
Inter-AS introduces techniques to establish MPLS VPNs across multiple
autonomous systems.
There are many options for:
- Exchanging VPN information
- BuildingVPN tunnels

In traditional service provider networks, MPLS VPN was mostly used inside one autonomous
system (AS). If customers did not use the same service provider in all branch offices, they
would not be able to establish an MPLS VPN.
Inter-AS introduces techniques to establish MPLS VPNs across multiple autonomous systems.
Service providers have to establish interconnection, exchange VPN information, and build VPN
tunnels.

3-60 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 An MPLS VPN tunnel is established across two service providers.

CustomerB
Site 2
CE4

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

In this example, two customers with two sites are connected to different service providers. An
MPLS VPN tunnel is established using an inter-AS connection between service providers.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-61

 ~- ..-
There are three options for configuring inter-AS:
- OptionA: back-to-back VRF
- Option B: single-hop MP-EBGP method
- Option C: multihop MP-EBGP between route reflectors
Option A is the simplest method.
Option C is the most scalable method.

There are three basic techniques for establishing inter-AS:

Option A: back-to-back VRF
The simplest method but not scalable
Option B: single-hop Multiprotocol Exterior Border Gateway Protocol (MP-EBGP)
Scalable method
Some routing overhead in Autonomous System Boundary Routers (ASBRs)
Option C: multihop MP-EBGP between route reflectors
Scalable method
End-to-end VPN between PE routers

3-62 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
esc Models
This topic describes different esc models.

o MPLS VPN is configured in backbone carrier.

o Customer carrier POP sites:
- Connected using Layer 3 MPLS VPN
- Run IGP and LDP with backbone carrier

Backbone
Carrier

~
MPLSVPN

MPL5VPN

Customer Customer
Site Site 2

C2012Ciscoandioritsafflllllte!l.A1I~ghts",served.

This esc implementation builds on the MPLS and LDP model. The resulting end-to-end LSP
enables the customer carrier to establish a peer relationship between its PE routers that are
supporting the end customer. It then enables the customer carrier to use MPLS VPNs to support
its end customers over an end-to-end VPN.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-63

 .
5
interface GigabitEthernetOjOjO/l
description Link PE-ASBR
vrf Customer carrier
ipv4 address-10.10.10.1 255.255.255.252
!
mpls Idp

interface GigabitEthernetOjOjO/l
!
router ospf 1 Backbone
address-family ipv4 Wlicast Carrier PE2
vrf Customer_carrier
area 0
interface GigabitEthernetOjOjO/l
!

interface GigabitEthernetOjOjO/l RR2

description Link PE-ASBR
ipv4 address 10.10.10.2 255.255.255.252
!

interface GigabitEthernetOjOjO/l
!
Customer
router ospf 1
Site
address-family ipv4 Wlicast
area 0
interface GigabitEthernetOjOjO/l
!

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

To configure CSC using IGP and LDP, you have to do the following:
Enable MPLS LDP for label exchange on a link connecting the backbone carrier and the
customer carrier
Configure IGP (OSPF in this example) to exchange routing information
It is important to have connectivity between route reflectors to establish an MP-BGP session.

3-64 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 o Backbone carrier establishes MPLS VPN for customer carrier.
o Customer carrier establishes MPLS VPN for end customers.

Configure an MP-IBGP session between

route reflector routers:
- Session between loopback interfaces

Configure an MP-IBGP session between PE routers:

- Session between loop back interfaces
- Send labels with customer carrier routes
- Override customer carrier AS number in AS path

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

MP-BGP is established between the route reflector routers of the customer carrier. When IBGP
is established between two routers, they ignore routes from same AS by default. You can
configure the PE router of the backbone carrier to override the AS number so that BGP routes
are accepted by IBGP. On routers using Cisco lOS XR Software, you can use the remove-
private-as command. On routers running Cisco lOS Software, the as-override command can
be used.
Backbone carriers should be able to send labels with IP prefixes. On lOS XR routers, the
address-family ipv4 labeled-unicast or address-family ipv6 labeled-unicast commands are
used. On lOS routers, you have to configure the send-label parameter for the BGP peer.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-65

 o When an IP packet enters the customer carrier VPN, an LDP label is
attached to it.
o When the packet arrives at the backbone carrier, another VPN label is
attached to it.

-- VPN

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.
VPN

In this example, a customer sends an IP packet to the service provider. A VPN label is attached
to the packet. This label is used to identify packets from the same VPN. Another LDP label is
attached to packet in order to forward the packet in the carrier network. When the packet enters
the backbone carrier network, another VPN label is attached to the packet. This label identifies
all packets from this provider.

3-66 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 o MPLS VPN is configured in backbone carrier.
o Customer carrier POP sites:
- Connected using Layer 3 MPLS VPN
- Run MP-EBGP with backbone carrier ASBR
- Use /32 loopback address for MP-IBGP sessions between route reflectors.
- On Cisco lOS XR routers, a static route should be configured on the backbone
carrier PE router pointing to the carrier ASBR router.
Backbone
Carrier

MP-IBGP

RR1

MPLSVP

Customer
sit!)
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

When configuring CSC using MP-BGP, LDP is not used for label distribution. MP-BGP can
exchange labels and route information between BGP peers.
If you are using routers with Cisco lOS XR Software, you have to configure a new address
family using the ipv4 labeled-unicast address-family or ipv6 labeled-unicast address-family
commands.
If you are using routers running Cisco lOS Software, send-label should be configured under
the BGP peer address family.
When you configure a BGP session between customer carrier POP sites, you should use /32
loopback addresses for source and destination IP addresses. If the IP address mask is not /32, it
can cause problems with label assignment on backbone carrier PE routers.
If you are using a router running lOS XR Software as the backbone carrier PE router, you have
to configure a static route toward the carrier ASBR router pointing to the physical interface that
connects both routers.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-67

 ~-----
interface GigabitEthernetO/O/O/l
description Link PE-ASBR
vrf Customer carrier
ipv4 address-lD.lD.lD.l 255.255.255.252

router static
vrf Customer_carrier
address-family ipv4 unicast
10.10.10.2/32 GigabitEthernetO/OjO/l

router bgp 64500 Backbone

vrf Customer_carrier PE1 Carrier PE2
rd 1:220
address-family ipv4 unicast

t
redistribute connected il.
allocate-label all
:
neighbor 10.10.10.2 ~
remote-as 64512
update- source GigabitEthernetO/ 0/0/1
address-family ipv4 unicast POP1
~ASBR1 ASBR2
route-policy pass in
route-policy pass out
as-override
next-hop-self
!
address-family ipv4 labeled-unicast
route-policy pass in
route-policy pass out
as-override Customer
next-hap-self
Si!!J
!

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

This example shows how to configure a BGP peer-facing customer carrier on the PE router of
the backbone carrier. The labeled-unicast address family is configured and the as-override
command is used to rewrite the local AS in the AS path.

When an IP packet enters the customer carrier VPN, an LDP label is

attached to it.
When the packet arrives at the backbone carrier, another VPN label is
attached to it.

i
- - - --
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.
VPN VPN

~
VPN VPN

~ ~

Data flow when using MP-BGP is similar to data flow when using LDP and IGP. First, a VPN
label is used to identify the customer VPN and a second VPN label is used to identify the
customer carrier VPN.

3-68 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Inter-AS
This topic describes the inter-AS method for interconnecting service provider networks.

ASBR routers are connected over multiple subinterfaces.

IGP runs between ASBR routers.

~
customerB

t!J RR1 CE2

Site 1
_,=.:o~
SP1
AsX ";'2

fIf~P-BGP
AsBR1
Multiple
IGP ~---- 5ubinterfaces
(

MP-BGP~
I sP2
Asy

I...RR2
CustomerB
Site 2
CE4

C2012Ciscoandioritsafflllllte!l.A1I~ghts",served.

You can configure inter-AS functionality using different techniques. The first is called Option
A, or the back-to-back VRF method.
ASBRs are interconnected using multiple interfaces or subinterfaces. Each interface or
subinterface is used to carry the traffic of its own VPN.
Each ASBR is acting as a PE router for its customers and a CE router for customers of other
service providers. Some IGP is used to exchange customer routing information between
ASBRs.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-69

 -.r-...._ _
ASBR needs to allocate a physical or logical link for each VPN.
Suitable when the number of VPNs is small
Not scalable
EachAS constructs its own VPN tunnel.
ASBRs act as CE routers for customers in an AS:
- ASBR needs to process routes of all VPN customers.

The back-to-back VRF inter-AS method is only suitable when two service providers have a
small number ofVPN tunnels. This method is easy to configure, but it is not scalable. Because
ASBRs act as CE routers for all VPN customers, these routers have to maintain routes from all
customers in their memory.

3-70 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 BGP is used to signal VPN labels between the AS boundary routers.
Higher scalability

~
customerB
Site 1
t!1 RR1 CE2 - ."-""-
SP1
AsX ff2

fI1 ~ MP-IBGP
AsBR1

MP-EBGP
(

MP-IBGP~
I sP2
Asy

m..RR2
CustomerB
Site 2
CE4

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The second method to configure inter-AS is called Option B, or the single-hop MP-EBGP
method.
Inside an AS, normal MPLS and BGP are used to transfer VPN information and construct the
LSP tunnel. Between autonomous systems, the single-hop MP-EBGP method is used to transfer
VPN information and construct the LSP tunnel.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-71

 Only one link is used between ASBRs.
Inter-AS link in the global table
Labels are exchanged between directly attached ASBRs.
Provides greater scalability
LSP tu nnel construction:
- Next-hop-selfmethod
ASBR announces itself as the next hop to the BGP neighbor.
New label is allocated
- Redistribute method
Routes to BG P peers are redistributed into IGP.

The single-hop MP-EBGP method provides more scalability than the back-to-back VRF
method. Only one link is configured between service providers. MP-BGP is used co exchange
routing and label information between directly connected routers.
To construct the LSP path, next-hop addresses should be reachable. Routes to BGP peers can
be redistributed to the provider IGP, or next-hap-self can be used by having the ASBR replace
the next hop with its IP address.

3-72 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Labeled IPv4 routes are redistributed by EBGP between neighboring
autonomous systems.
BGP is used for label distribution.

~
customerB
Site 1
CE2-.o,--

PE~

MP-IBGP

CustomerB
Site 2
CE4
1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The third method to configure inter-AS is called Option C, or multihop MP-EBGP.

Because BGP only requires the TCP connection to form a BGP neighbor and transfer route
information, this third method transfers VPN route information between the source and
destination PEs directly over multihop MP-EBGP, and constructs a public network LSP tunnel
between the source and destination PEs. When there is route reflector in service, a provider
network BGP session can be established between route reflectors.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-73

 ASBRs do not have VPNv4 routes and label information.
MP-EBGP peering between route reflectors in different autonomous
systems.
BGP is used for label distribution betweenASBRs.
End-to-end LSP is required from ingress PE to egress PE.
You can use a route map or route policy to filter the distribution of MPLS
labels between routers.

To use the multihop MP-EBGP method, end-to-end LSP is required from the ingress PE (or
route reflector) to the egress PE (or route reflector). This method is highly scalable, because
there is no route overhead on ASBRs.
You can use a route map or route policy to filter the distribution of MPLS labels between
routers.

3-74 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary points that were discussed in this lesson.

The two basic MPLS interdomain solutions are esc and inter-AS.
esc is a hierarchical method for interconnecting service providers.
Inter-AS is a peer-to-peermethod for interconnecting service providers.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-75

3-76 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Summary
This topic summarizes the primary points that were discussed in this module.

3
Overlapping VPNs are used to provide connectivity between certain
segments of two separate VPNs.
- Service providers use central services to provide access to common
infrastructure and services.
Service providers can use the same infrastructure to provide MPLS
service and Internet access to customers.
- Customers can have multisite or centralized Internet access.
Inter-AS and esc are two methods for interconnecting service
providers.

02012CilCO.ndlor~.afl'ili_.Allrighlol . . .rv"'.

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-77

3-78 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Self-Check
Use the questions here to review what you learned in this module. The correct answers and
solutions are found in the Module Self-Check Answer Key.
QI) Why do you need a selective VRF import command? (Source: Implementing Complex
MPLS Layer 3 VPNs)

Q2) Why do you need a selective VRF export command? (Source: Implementing Complex
MPLS Layer 3 VPNs)

Q3) Who are the typical users of overlapping VPNs? (Source: Implementing Complex
MPLS Layer 3 VPNs)

Q4) What are the connectivity requirements for overlapping VPNs? (Source:
Implementing Complex MPLS Layer 3 VPNs)

Q5) What are the typical usages for a central services VPN topology? (Source:
Implementing Complex MPLS Layer 3 VPNs

Q6) Why do you need the managed CE routers service? (Source: Implementing Complex
MPLS Layer 3 VPNs)

Q7) What is the main difference between the managed CE routers service and the typical
central services VPN topology? (Source: Implementing Complex MPLS Layer 3
VPNs)

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-79

 Q8) What are the two major design models for implementing Internet access and MPLS
VPNs? (Choose two.) (Source: Implementing Internet Access and MPLS Layer 3
VPNs)
A) using a separate VPN for Internet access
B) using global route leaking
C) using global routing on PE routers
D) using default routes
Q9) What is the recommended implementation option for using global routing to provide
Internet access? (Source: Implementing Internet Access and MPLS Layer 3 VPNs)
A) The separate subinterfaces are placed in separate VRFs.
B) The Internet subinterface is not placed in a VRF.
C) One interface for the Internet and the MPLS VPN is recommended.
D) Static routes are always needed for Internet access.
Q10) Which two are drawbacks of Internet access that is implemented as a separate VPN?
(Choose two.) (Source: Implementing Internet Access and MPLS Layer 3 VPNs)
A) Full Internet routing cannot be carried inside a VRF.
B) The customer cannot receive full Internet routes.
C) Default routing toward the Internet gateways must be used, potentially
resulting in suboptimal routing.
D) The customer must use default routes.
Qll) Which two of these are true about the CSC model? (Choose two.) (Source: Introducing
MPLS Interdomain Solutions)
A) CSC is a peer-to-peer model.
B) CSC is a hierarchical model.
C) Multiple customer carriers can be connected to a single CSC backbone.
D) End customer routes are inserted in the CSC backbone.
Q12) Which two protocols are used to exchange label and route information between the
backbone and customer carriers? (Choose two.) (Source: Introducing MPLS
Interdomain Solutions)
A) CDP
B) OSPF and LDP
C) MP-BGP
D) NTP
Q13) Which two of these are true about the inter-AS model? (Choose two.) (Source:
Introducing MPLS Interdomain Solutions)
A) Inter-AS is a peer-to-peer model.
B) Inter-AS is a hierarchical model.
C) The back-to-back VRF method is suitable for small environments but is not a
scalable solution.
D) In the multihop EBGP method, the ASBR router contains customer routes.
Q14) What are the names of the three basic options for configuring inter-AS? (Source:
Introducing MPLS Interdomain Solutions)

3-80 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Self-Check Answer Key
Q l) A selective VRF import command allows you to select routes to be imported into a VRF based on criteria
other than the VRF R T.
Q2) A selective VRF export command allows you to attach specific RTs to a subset of routes exported from a
VRF. (By default, the same RTs get attached to all exported routes.)
Q3) Companies that use MPLS VPNs to implement both intranet and extranet services, or a security-conscious
company that wants to limit visibility between different departments in the organization
Q4) Selected sites in a VPN can communicate only with sites within their VPN. Other selected sites can
communicate with sites in their VPN and selected sites in a second VPN.
Q5) In solutions where some sites (server sites) can communicate with all other sites, but all the other sites
(client sites) can communicate only with the server sites
Q6) If the service provider is managing the customer routers, it is convenient to have a central point that has
access to all CE routers but not to the other destinations at customer sites.
Q7) The VRF and RD design is similar to that of a central services VPN. The managed CE routers service
combines a service VPN and simple VPN topology like the central services VPN. However, the route
export statement uses a route policy to limit the exported addresses to the loopback address of the managed
routers.
Q8) A, C
Q9) B
QlO) A, C
Qll) B, C
Ql2) B, C
Ql3) A, C
Ql4) Option A, or back-to-back VRF; Option B, or the single-hop MP-EBGP method; and Option C, or the
multihop MP-EBGP method

2012 Cisco Systems, Inc. Complex MPLS Layer 3 VPNs 3-81

3-82 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Module 41

Layer 2 VPNs and Ethernet

Services
Overview
Virtual Private Wire Service (VPWS) enables geographically separated sites to be
interconnected over a virtual point-to-point Layer 2 circuit. The virtual connection can cross a
Multiprotocol Label Switching (MPLS) or IP network. Virtual Private LAN Services (VPLS)
enable remote LAN segments to be linked as a single bridged domain over an MPLS network.
The full functions of the traditional LAN, such as MAC address learning, aging, and switching,
are emulated across all the remotely connected LAN segments that are part of a single bridged
domain.
A service provider can offer VPLS service to multiple customers over the MPLS network by
defining different bridged domains for different customers. Packets from one bridged domain
are never carried over or delivered to another bridged domain, thus ensuring the privacy of the
LAN service. VPLS transports Ethernet 802.3, VLAN 802.IQ, and VLAN-in-VLAN (QinQ)
traffic across multiple sites that belong to the same Layer 2 broadcast domain. VPLS offers
simple VLAN services that include flooding broadcast, multicast, and unknown unicast frames
that are received on a bridge. The VPLS solution requires a full mesh of pseudowires (PWs)
that are established among provider edge (PE) routers. The VPLS implementation is based on
Label Distribution Protocol (LDP)-based PW signaling.
This module describes the VPWS and VPLS technology and implementation on Cisco
platforms.

Module Objectives
Upon completing this module, you will be able to describe Layer 2 VPNs and Ethernet
services. You will be able to meet these objectives:
Describe Layer 2 VPNs that are available in the MPLS and IP core
Describe AToM
Describe Ethernet services that are used in the service provider network
4-2 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 11

Introducing Layer 2 VPNs

Overview
There is an ever-increasing demand for the transport of Layer 2 and Layer 3 over a common
backbone. This lesson introduces the Virtual Private Wire Service (VPWS) and Virtual Private
LAN Services (VPLS). VPWS offers point-to-point virtual connections, while VPLS provides
LAN-similar multipoint connectivity. A subset ofVPWS is Any Transport over MPLS
(AToM). AToM allows a Multiprotocol Label Switching (MPLS) network to provide end-to-
end transport for Layer 2 frames and cells. It provides support for Ethernet, PPP, High-Level
Data Link Control (HDLC), Frame Relay, and ATM.

Objectives
Upon completing this lesson, you will be able to describe Layer 2 VPNs that are available in
the MPLS and IP core. You will be able to meet this objective:
Explain Layer 2 VPN services that are available with IP and MPLS core
Layer 2 VPN Overview
This topic explains Layer 2 VPN services that are available with IP and MPLS core.

New service opportunities:

- Virtual leased-line service
- Offer "PVC-like" Layer 2-based service
Reduced cost: consolidate multiple core technologies
into a single packet-based network infrastructure.
Simplify services: Layer 2 transport provides options for service
providers who need to provide Layer 2 connectivity and maintain
customer autonomy.
Protect existing investments: Extend customer access to existing Layer
2 networks without deploying a new separate infrastructure.
Feature support: Through the use of Cisco lOS features
such as IPsec, QoS, and traffic engineering, Layer 2 transport
can be tailored to meet customer requirements.

C2012Ci'"""8rd'oritstrffili*'".A1lrigl'ts......,'......

Layer 2 VPN technologies offer a range of benefits to service providers and enterprises:
New service opportunities: The virtual leased lines offer connectivity service that
resembles the traditional costly permanent virtual circuits in Frame Relay or ATM
environments.
Cost savings: The consolidation of multiple core technologies into a single packet-based
network infrastructure lowers the overall cost of system installation and maintenance.
Simple connectivity model: Layer 2 transport provides options for service providers that
need to provide Layer 2 connectivity and maintain customer autonomy.
Investment protection: Service providers can extend customer access to existing Layer 2
networks without deploying a new separate infrastructure.
Feature support: Through the use of Cisco lOS and lOS XR features such as IPsec,
quality of service (QoS), and traffic engineering, Layer 2 transport can be tailored to meet
customer requirements.

4-4 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Attachmentcircuit (e.g., Frame Relay OLCI, PPP) mapped to emulated
VC
Pseudowire: Connection between two PE devices that connect two
attach ment circuits
Transport: MPLS or IP (L2TPv3)

.~
~ AC

MPLS

The Layer 2 VPN solution encompasses three main elements:

An attachment circuit is the circuit or link directly connected to the provider edge (PE)
system that is virtually extended to the other end of the provider cloud. It can represent a
Frame Relay data link connection identifier (DLCI), PPP and HDLC link, ATM virtual
path identifier (VPI) and virtual channel identifier (VCI) pair, Ethernet port, or VLAN. The
attachment circuit is mapped to the emulated virtual circuit (VC) for transport through the
service provider core.
Pseudowire (PW) is a point-to-point connection over a packet-switching network. The PW
emulates the operation of a "transparent wire" by linking two distant attachment circuits
attached to two different PE routers.
Transport infrastructure is the third element. The transport network can either be MPLS-
enabled, and thus capable of supporting any Layer 2 topology (point-to-point, point-to-
multipoint, or multipoint-to-multipoint), or IP-based. In the IP core, the Layer 2 Tunneling
Protocol (L2TP) protocol supports point-to-point connections only.
Frames are received on an ingress interface by the ingress PE router. At this point, the frame is
a raw Layer 2 frame. In the case of MPLS transport, the ingress PE router encapsulates it into
MPLS and tunnels it across the backbone to the egress PE router. The egress PE router
decapsulates the packet and reproduces the raw Layer 2 frame on the egress interface.
The frames are carried across the MPLS backbone using a label stack of two labels. The top
label is used to propagate the packet from the ingress PE router to the correct egress PE router.
The second label is used by the egress PE router to forward out the packet on the correct
interface. This process is somewhat similar to an MPLS VPN, where the egress PE router uses
a VPN label. The top label is called the tunnel label. This name indicates that its use is to tunnel
the packet across the MPLS backbone to the egress PE router. The second label is called the
VC label. The name indicates that its use is to map the packet to an outgoing VC or link.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-5
 Full Efficient Large Scale Intelligent Multiservice Intelligent Efficient Full
Service CPE Access Aggregation Edge Core Edge Access Service CPE
U-PE PE-AGG N-PE P N-PE U-PE

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Carrier Ethernet environments are typically designed and deployed according to the
architecture seen in the figure. It illustrates the various layers: the multiservice core, the
network edge with intelligent services, aggregation of large amounts of access circuits, the
access layer, and the customer equipment. In many cases, a specific carrier or Metro Ethernet
solution may not contain all ofthese layers. In fact, in some cases the architectural functions
can be merged into a single layer. For example, various combinations of network technologies
and topologies can be formed to deliver Ethernet services without passing through a core
network. In this context, these network technology and topology combinations can be viewed
as separate from the interconnecting core network, and are hence referred to as Metro Ethernet
islands (or simply islands).

4-6 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 E-Line:
- Ethernet private lines
- Ethernet virtual private lines
- Ethernet Internet access
E-LAN:
- Multipoint Layer 2 VPNs
- Transparent LAN service
- Foundation for IPTV and multicast networks
E-Tree:
- Also known as rooted multipoint
- Leaves can communicate with one or more
roots
- Leaves do not communicate with other
leaves
Targeted at multi host separation
Enabler for mobile backhaul and triple-play
infrastructure

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

To bring about the industry acceptance of Ethernet services, it was necessary to clarify and
standardize the service range. Recognizing this requirement, the industry created the Metro
Ethernet Forum (MEF), which played a key role in defining the three major services:
E-Line: a service connecting two customer Ethernet ports over a WAN. This service is
further subdivided into Ethernet private lines (EPLs), Ethernet virtual private lines
(EVPLs), and Ethernet Internet access (EIA).
E-LAN: a multipoint service connecting a set of customer endpoints, giving the appearance
to the customer of a bridged Ethernet network connecting the sites. This transparent LAN
service is often referred to as a multipoint Layer 2 VPN. It lays the foundation for IPTV
and multicasting applications.
E- Tree: a multipoint service connecting one or more roots and a set of leaves, but
preventing interleaf communication. This service is also known as rooted multipoint.
Specifically, the leaves can communicate with one or more roots, but not with other leaves.
The service provides an ideal mechanism for multihost separation. It is considered a major
enabler for mobile backhaul and triple-play infrastructure.
All these services provide standard definitions of such characteristics as bandwidth, resilience,
and service multiplexing, allowing customers to compare service offerings and facilitating
service level agreements (SLAs).

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-7
 Metro Ethernet Forum

Ethernet private
IElF (MPLS)
III
QinQ, .1ad
Cisco service
name
Ethernet Wire
E-Line (point- line (EPL) Service (EWS)
Virtual Private
to-point) Wire Service
Ethernet virtual .1Q Ethernet Relay
(VPWS)
private line Service (ERS)
(EVPL)
Transparent LAN QinQ, .1ad Ethernet Multipoint
Service (TLS) Service (EMS)
Virtual Private
E-LAN
LAN Service
(multipoint) Ethernet Virtual .1Q Ethernet Relay
(VPLS)
Connection Multipoint Service
Service (EVCS) (ERMS)

MEF does not define only the three major categories (E-Line, E-LAN, and E-Tree), but many
variants of them. One classification of the variants is based on the Ethernet virtual circuit
(EVe) mode: port- or VLAN-based.
In addition, the terms E-line, E-LAN, and E-Tree are not the only ones used in the industry.
IETF refers to these same services as VPWS and VPLS. The IETF naming is used
predominantly throughout this course.

4-8 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 IEEE 802.1ad:
- Formal name of 802.1QinQ
IEEE 802.1ah:
- Also known as provider backbone bridges (PBB), or MAC in MAC
- Removeslimrtations ofVPLS
Flat MAC topology
Number of VLANs
- Scales to large service provider environments
IEEE 802.1ag:
- Connectivity Fault Management (CFM)
- Protocols and practices for Operations, Administration, and Maintenance (OAM)
- Three protocols:
Continuity Check Protocol
Link Trace
Loop-back

Many standardization bodies have issued standards that define basic functions and
enhancements of the Carrier Ethernet architecture. IEEE plays a major role in this field and
contributed many important standards, such as the following:
IEEE 802.1 ad, which is the formal name of the 802.1 QinQ standard that allows Ethernet
frame encapsulation in multiple VLAN tags.
IEEE 802.1ah, also known as provider backbone bridges (PBBs), or MAC in MAC. This
technology addresses the limitations of VPLS, such as flat MAC topology or limited
number ofVLANs. It scales to large service provider environments.
IEEE 802.lag, also referred to as Connectivity Fault Management (CFM). This set of
recommendations defines protocols and practices for Ethernet Operations, Administration,
and Maintenance (OAM). Specifically, it comprises three protocols: Continuity Check
Protocol, Link Trace, and Loop-back.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-9
 r------------1 Layer 2 VPN Models

VPWS

Like-to-like
Any-to-eny
P2P

PPPIHDLC PPP/HDLC

ATM AAL5/Gen ATM AAL5/Cell

Ethernet Ethernet

Frame Relay Frame Relay

Layer 2 VPNs are grouped into three main categories: local switching that serves directly
connected links, methods for transport in the MPLS core, and transport solutions for IP core.
MPLS core transport is further subdivided into VPWS and VPLS).
VPWS is a point-to-point technology. MPLS-based VPWS is called Any Transport over MPLS
(AToM) and supports connections between the same interface types (like-to-like), and between
different interface types (any-to-any). The supported attached interface types include Ethernet,
Frame Relay, and ATM, including ATM adaptation layer 5 (AAL5), PPP, and HDLC.
VPLS offers point-to-multipoint and multipoint-to-multipoint connectivity. It leverages MPLS
as the transport infrastructure.
IP core transport uses the Layer 2 Tunneling Protocol version 3 (L2TPv3), and supports only
point-to-point connections. The available encapsulations include Ethernet, Frame Relay, and
ATM, including AAL5, PPP, and HDLC. They can be linked in any-to-any fashion.

4-10 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Single infrastructure for both IP and traditional services
- Service providers:
Move legacy ATM and Frame Relay traffic to the MPLS or IP core without
servi ce interrupti on
- E nterpri ses:
Optimize data center solution with WAN or M PLS transport
Improve high availability
New Layer 2 tunneling services
- Customercan have its own routing, QoS policy, and so on
A migration step toward IP and MPLS VPN

Layer 2 VPN allows both service providers and enterprise to build a single infrastructure for
both IP and traditional services. The service providers can migrate their existing ATM and
Frame Relay traffic to MPLS or IP core without interrupting the customer service. The
enterprises can leverage the extended Layer 2 domain to optimize the data center solution. The
MPLS transport enables a host of additional high availability extensions.
The Layer 2 VPN is a tunneling technology that provides logical separation between the
customer and service provider domains, including segmentation of routing, QoS policy, and
others.
Layer 2 VPNs also represent an easy-to-implement migration step toward a Layer 3 IP and
MPLS VPN system.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-11
 Appearance of CE-to-CE native service (transparent service provider
network)
Negotiation ofVC labels or session IDs
Signaling and intelWorking with native services (e.g., Frame Relay LMI)
Discovery of other PE VPN members (MP-BGP, LOP)

Native service:
: '(FrameRel,,'
Pseudowire
j+.---------------+ .
Native service
(Frame Relay,
..
! Ethernet, , Ethernet,
HOLC, etc.) Neighbor disODvery HOLC, etc.)

CE
. . . . . . . _.......................- .HJI--'I
LOP signalling

A protocol is required between the PE routers so that they can exchange the VC information.
In the case of MPLS transport, the Label Distribution Protocol (LDP) is used for this purpose.
A directed multihop LDP session is established between the PE routers. The egress PE router
sends an LDP message in which it indicates the label value to use for a virtual circuit
forwarding equivalence class (VC FEC). That label value is then used by the ingress PE router
as the second label in the label stack that is imposed to the frames of the indicated VC FEC.
In the case of IP-based transport, the L2TPv3 session exchanges session parameters and not
labels.
The figure shows a directed multihop LDP session between the ingress and egress PE routers
that is used to exchange the VC label. Any ingress-egress PE router pair will need such an LDP
seSSIOn.
The control session is also responsible for providing interworking capabilities with native
services, such as Frame Relay Local Management Interface (LMI). The neighbors can be either
statically configured or auto-discovered using Multiprotocol Border Gateway Protocol (MP-
BGP) extensions or LDP.

4-12 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEOGE) v1.0 2012 Cisco Systems, Inc.
 Transport header: IPv4
Tunnel header.
----
- 4-byte session ID with optional8-byte cookie
- Signaled or statically configured
Layer 2 PDU: Layer 2 specific sublayerwith payload (CE Layer 2 PDU)

, Native service:
; c
! (FR, Ethernet,
~
Pseudowire
:---......;..;;.,;;",;;;;;;;.;,;,;;,;;...---..:...._1
:
! +==..;.;",;.~

i HDLe, elc,

CE
...._ _._ t!J _ - .

L2TPv3 encap

L2TPv3 is used to transport Layer 2 frames over pure IP networks. The entire L2TP packet,
including payload and L2TP header, is sent within a UDP datagram. Traditionally, L2TP has
been used to carry PPP sessions within an L2TP tunnel. L2TPv3 provides additional security
features, improved encapsulation, and the ability to carry data links other than just PPP over an
IP network (for example, Frame Relay, Ethernet, ATM, and others).
L2TP overhead includes the transport IP header (20 bytes) and an L2TP header of variable
length. The only mandatory field in the L2TP header is the session ID (4 bytes). Optional fields
are cookie (8 bytes) and control word. The payload of the L2TPv3 packet is the original Layer
2 protocol data unit (PDU).

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-13
 - Signaled via LOP or M PLS-TE
-
Transport header: MPLS label (4-byte)

- Unidirectional path to egress PE

Tunnel header. VC Label
- Signaled via directed LOP
Layer 2 PDU: Control word with customer payload (may not include
entire Layer 2 h8::H 'lAri

.-
, Native service: Pseudowire Native service
; c ~ :-----...,;...;.,;;,.;=;,;.;;,;;-----...,;.-_+I! .. ~
! (FR, Ethernet, : (FR, Ethernel,
i HDLC, elc, DLC, etc,)

~~j~.
CE
...._ _._ t8 _ - . CE

MPLS headers

MPLS transport is based on the same mechanism that you examined for Layer 3 MPLS VPNs.
Hop-by-hop LDP signals a unidirectional path to the egress PE router. A directed LDP session
exchanges the VC label that serves as the inner label in the MPLS packet. In general, the use of
control words is optional, and configurable. Some transport types, such as Frame Relay Layer 2
VPN, require the use of control words.
Depending on the frame encapsulation, some fields of the original frame, such as checksums,
may be stripped before MPLS encapsulation.

4-14 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 VPWS
- Point-to-pointLayer2 connections
- No MAC learning
- Two transport methods:
L2TPv3
AToM
- Example: Ethernet over M PLS
VPLS
- M ultipoi nt Layer 2 connections
- Collectionof PWs tied together by a VFI
- MAC addresses learned on VFI
- Traffic forwarding based on destination MAC addresses
- Allows hierarchical topologies (H-VPLS)

Layer 2 transport services are categorized into two major classes:

VPWS: This service type is used to deploy point-to-point Layer 2 connections. It does not
involve MAC learning capabilities. It encompasses two transport methods:
L2TPv3
AToM. The most common example of AToM is Ethernet over MPLS (EoMPLS).
VPLS. This technology supports multipoint Layer 2 connections by grouping a collection
ofPWs terminated on a PE router in a Virtual Forwarding Interface (VFI). The VFI
represents a virtual extension of the physical circuit attached to the PE system. The VFI
resembles a switch that is capable of learning MAC addresses and forwards traffic based on
its MAC address table. A VPLS can connect thousands of PEs into a single VLAN and is
therefore subject to scalability constraints. To improve the scalability of the solution,
hierarchical VPLS (H-VPLS) topologies enable a two-tier deployment of the PE devices.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-15
 Flavor of AToM (attachment circuit: Ethernet; transport: MPLS)
Attachmentcircuitcan be based on:
- Port (VC label type Ox0005)
- VLAN (VC label type Ox0004)
- Ethernetflow point

Customer A Customer A

Physical ~ Switr:h

topology:
PE
Site 2

Logical
topology: BPDUs, VTP Messages

EoMPLS is the most ubiquitous example of AToM, which transports Ethernet frames over an
MPLS infrastructure. The virtual link bridging the Ethernet segments at both ends is transparent
to shortest path tree (SPT) bridge protocol data units (BPDUs), Virtual Terminal Protocol
(VTP) packets, and other control messages.
The attachment circuit can be the Ethernet port or 802.1 Q subinterface (VLAN). For each
attachment circuit, LDP signals a different VC type via the targeted LDP session. VC type 5 is
used for Ethernet port mode and VC type 4 is used for Ethernet VLAN mode.
In Ethernet port mode, both ends of PW are connected to Ethernet ports. In this mode, the port
is tunneled over PW or, using local switching (also known as attachment circuit-to-attachment
circuit cross-connect) switches packets or frames from one attachment circuit to another
attached to the same PE node. In Ethernet port mode, the PW is always a type 5 virtual
connection. On the ingress PE, the network service provider passes the packets to the PW
termination point, adds the MPLS labels to the packets, and sends the packets over the PW. In
Ethernet port mode, a VLAN header mayor may not be present in the frame. In any case, the
PE router carries the frame transparently. This allows an Ethernet trunk to be carried over a
single PW.
VLAN mode provides Ethernet VLAN-to-VLAN connectivity. In VLAN mode, each VLAN
on a customer-end to provider-end link can be configured as a separate Layer 2 VPN
connection, using either virtual connection type 4 or type 5.
Virtual connection type 5 is the default mode. In type 4 virtual connections, on the ingress PE,
the VLAN tag maps to a particular PW and the packet is placed on the PW with the VLAN tag
untouched.
In Type 5 virtual connections, on the ingress PE that is receiving packets from the customer
edge (CE), the network service provider strips off the customer edge VLAN tag before placing
the packets on the PW. On the egress PE, the network service provider pushes the VLAN tag
onto the protocol stack before it sends the packet to the CEo

4-16 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Allows different Layer 2 encapsulations at opposite ends
Extension of "I ike-to-like" to "any-to-any" concept

(FR, Ethernet, (FR, Ethernet,

! HDLC, PPP,ATM)~ ..;.p.;.S,;;,;eU;;;;d.;.OW;,;,;i;;;;re _ HDLC, PPP, ATM)

i :

CE
PE CE

MPLS headers

Ethernet Ethernet
Frame Relay
Frame Relay
PPPA-iOLC PPP/HOLC
ATM ATM

Layer 2 VPN interworking allows you to connect disparate attachment circuits. Cisco routers
support these any-to-any combinations:
Ethernet or VLAN to AIM AAL5 interworking
Ethernet or VLAN to Frame Relay interworking
Ethernet or VLAN to PPP interworking
Ethernet to VLAN interworking
Frame Relay to AIM AAL5 interworking
Frame Relay to PPP interworking
Ethernet or VLAN to AIM VPl and VCl interworking

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-17
 The Layer 2 VPN interworking function is implemented in two modes:
- Bridged interworking mode:
Ethernetframes are extracted from the attachment circuit.
Non-Ethernet frames on attachment ci rcuit are dropped.
VLAN tag removed
- Routed interworking mode:
IPpackets are extracted from the attachment ci rcuit.
FrameswithoutlP packets are dropped.

The L2VPN interworking function is implemented in two modes:

Bridged interworking mode
Routed interworking mode
In bridged interworking mode, Ethernet frames are extracted from the attachment circuit and
sent over the PW. Attachment circuit frames that are not Ethernet are dropped. In the case of a
VLAN, the VLAN tag is removed, leaving an untagged Ethernet frame. This interworking
functionality is implemented by configuring the interworking ethernet command under the
pseudowire class configuration mode.
In routed interworking, IP packets are extracted from the attachment circuit and sent over the
PW. Attachment circuit frames are dropped if they do not contain the IPv4 packets. This
interworking functionality is implemented by configuring the interworking ip command under
the pseudowire class configuration mode.

4-18 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 hos tname PEl hos tname PE2

pseudowire-c1ass Eth-VLAN pseudowire-c1ass Eth-VLAN

encapsulation mp1s encapsulation mp1s
interworking ethernet interworking ethernet

interface EthernetO/1 interface EthernetO/1.10

no ip address no ip address
xconnect 10.10.10.100 100 encapsulation encapsulation 802.1Q 10
mp1s pw-c1ass Eth-VLAN xconnect 10.10.10.101 100 encapsulation
! mp1s pw-c1ass Eth-VLAN

: I Ethernet l' _---......;.P~s..;.e..;.ud;;.;o~w..;.ire..;....-----_,

: ~,_8_0_2._1Q
__

4
hos tname CE1
PE1

MPLS

hos tname CE2

interface EthernetO/O interface EthernetO/0.10

ip address 192.168.10.1 255.255.255.0 encapsulation dot1Q 10
ip address 192.168.10.2 255.255.255.0

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

The steps to configure Ethernet to VLAN interworking between CEI and CE2 are as follows:
Step 1 Define pseudowire class on PE routers
In this step, a pseudowire class called Eth-VLAN is defined on the PEl and PE2 routers. This
class configures the PW between the PE routers PEl and PE2. Ensure that the parameters ofthe
pseudowire class are the same on both PEs to enable PW establishment. The example shows
that AToM encapsulation (encapsulation mpls) and bridged interworking mode (interworking
Ethernet) will be used by the pseudowire class on the PE routers PEl and PE2.
Step 2 Define AToM VC to transport Layer 2 frames
In this step, use the xconnect statement to define the AToM VC to carry the Layer 2 frames
from CEI to CE2, and vice versa. Associate the pseudowire class defined in Step 1 with the
AToMVC.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-19
 End-to-end architecture
Layer 2 multipoint Ethernet service:
- MPLS transport (not L2TPv3)
- Virtual bridges linked with PWs
Service provider emulates an IEEE Ethernet bridge network
Same data plane as EoMPLS (point-to-point)

VPLS is an architecture

PE

CE
C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

The purpose ofVPLS is to provide a private multipoint LAN-type Ethernet connectivity

service. VPLS emulates a LAN segment over an MPLS backbone across PWs or virtual circuits
(VCs). VPLS creates one or more LANs for each customer who is using the service from the
service provider. Each LAN is completely separate from the other emulated LAN segments.
When a customer with different Ethernet sites connects to an MPLS backbone where VPLS is
deployed, it appears as if all the sites are interconnected through a virtual Ethernet switch.
For each VPLS, the PE routers are fully meshed with PWs. A PE receiving a frame from
another PE can identify which VPLS the frame belongs to, based on a PW label or VC label.
As far as each customer is concerned, an Ethernet frame that is sent into the service provider
network is delivered to the correct site(s) based on the destination MAC address. It is the task
of each PE router to inspect the destination MAC address of each frame arriving from a locally
attached site and to forward it to the appropriate destination site. This destination site may be
attached to the same PE on a different port or a remote PE. If the destination site is attached to
the same PE, the PE locally switches the frame to the correct port. If the destination site is
attached to a remote PE, the ingress PE must forward the frame to the appropriate PW to the
remote PE. This means that the ingress PE needs to know which egress PE to send the frame to.
There are two ways in which this can be accomplished. One is to have a control plane signaling
to carry information about MAC addresses between PEs; another is to have a scheme based on
MAC address learning. VPLS takes the latter approach by having each PE take the
responsibility for learning which remote PE is associated with a given MAC address. This way,
an ingress PE simply needs to identify which frames need to be sent to egress PEs, and egress
PEs take care of identifying which local ports to forward the packet to. By inspecting the source
MAC address of the frame arriving on a port, whether an actual local port or a PW from a
remote PE, and by creating a corresponding entry in the forwarding table, the PE learns where
to send future frames with that destination MAC address.
If Ethernet switches are used as CE devices and connected to PE routers, the PEs need to learn
the MAC addresses of individual hosts attached to the switches. So, if a host is plugged into the
office network served by a switch as a CE, the effect will be felt by all PEs. Thus, for a large
deployment, it is better to use routers as CEs than switches.

4-20 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 It is virtual because multiple instances of this
service share the same physical infrastructure.
It is private because each instance of the service
is independent and isolated from others.
It is LAN service because it emulates Layer 2 multipoint connectivity
between subscribers.
Benefits:
- Customers have full operational control over their routing neighbors.
- Privacy of add ressing space-no sharing with the carrier network.
- Customer has a choice of usi ng any routing protocol, includi ng non-IP.
- Customers can use an Ethernet switch instead of a router as the CPE.

VPLS has become a very attractive technology over the past few years with the advent of
MPLS. The reason for this is that some enterprises are very reluctant to relinquish the routing
control of their network to the service provider, and they desire Layer 2 VPN services with
multipoint connectivity. VPLS allows service providers to deploy carrier-class service over
Ethernet and MPLS-based networks in a reliable and flexible way.
The term implies these characteristics:
It is "virtual" because multiple instances of this service share the same physical
infrastructure.
It is "private" because each instance of the service is independent from the others.
It is "LAN service" because it emulates Layer 2 multipoint connectivity between
subscribers.

The main VPLS benefits include the following:

Customers have full operational control over their routing neighbors.
Privacy of addressing space means that addresses of the carrier infrastructure are
completely isolated.
Customers have a choice of using any routing protocol, including non-IP protocols such as
Intermediate System-to-Intermediate System (IS-IS).
Customers can use an Ethernet switch instead of a router as the customer premises
equipment (CPE) device.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-21
 Scales multipoint Layer 2 services:
- 16 million service IDs
Customer demarcation
MAC hiding
Flooding elimination
VPN aggregation

Provider Backbone Bridge

802.1 ad Inte~s/ _ or~k ~8
...N_e_tw.... ..(. ....
02....1_a_h)_"\,::--

Provider Bridge
Network (802.1ad)

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

Because of the lack of separation of customer networks from the carrier network (control
plane), lack of scalability (limit of 4094 VLAN IDs), and lack of end-to-end QoS needed to
achieve connection-oriented, carrier-grade Ethernet services, the standard known as provider
backbone bridges (PBBs), or 802.lah, was developed.
802.lad provides stacking ofVLAN IDs and allow separation of the customer VLAN ID from
the service provider VLAN ID.
But since the control plane operates at the MAC layer and the goal is to provide separation of
the customer control plane from the service provider control plane, one way is to simply
"stack" the MAC addresses in a similar manner. This "stack" MAC approach is defined in the
802.lah PBB standard, which is also referred to as MAC in MAC.
PBB is a set of architecture and protocols for routing over a provider network, allowing the
interconnection of multiple provider bridge networks without losing the individually defined
VLANs of each customer. It provides a further enhancement over QinQ tunneling to support
even larger Ethernet deployments. QinQ does not offer true separation of customer and
provider domains, but is merely a way to overcome the limitations on the VLAN identifier
space.
The idea of PBB is to offer complete separation of customer and provider domains. It addresses
the constraints of 802.lad, such as having too little control on the MAC addresses, since QinQ
forwarding is still based on the customer destination addresses. PBB eliminates flooding from
the provider infrastructure and allows an efficient VPN aggregation.
802.lad and 802.lah can still be used hand-in-hand, as shown in the figure. QinQ is commonly
used in the edge network, while PBB is deployed in the core.
PBB defines a new Ethernet header. The main components ofthe header are as follows:
Backbone component that has:
Backbone destination address (B-DA) (6 bytes)
Backbone source address (B-SA) 6 bytes)

4-22 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 EtherType Ox88A8 (2 bytes)
Backbone VLAN tag (B-TAG) and backbone VLAN ID (B-VID) (2 bytes); this is
the backbone VLAN indicator
Service encapsulation that has:
EtherType Ox88E7 (2 bytes)
Flags that contain priority, drop eligible indicator (DEI), and No Customer Address
indication (for example, OAM frames).
Service instance VLAN ID (I-SID) (3 bytes)
Original customer frame
Customer source address (6 bytes)
Customer destination address (6 bytes)
EtherType Ox81 00 (2 bytes)
Customer VLAN identifier (2 bytes)
EtherType (e.g. Ox0800)
Customer payload
PBB defines a 48-bit B-DA and 48-bit B-SA to indicate the backbone source and destination
MAC addresses. It also defines a 12-bit B-VID and 24-bit I-SID. The bridges in the PBB
domain switch based on the B-VID and B-DA values, which contain 60 bits total. Bridges learn
based on the B-SA and ingress port value and hence are completely unaware of the customer
MAC addresses. I-SID allows for distinguishing the services within a PBB domain.

Note Detailed information on PBB implementations is beyond the scope of this class.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-23
Summary
This topic summarizes the primary point that was discussed in this lesson.

Layer 2 transport services are classified as VPWS or VPLS.

4-24 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 21

Introducing AToM
Overview
Any Transport over MPLS (AToM) is a subset of Virtual Private Wire Service (VPWS) that
provides point-to-point virtual connections. AToM allows an MPLS network to provide end-to-
end transport for Layer 2 frames and cells. It provides support for Ethernet, PPP, High-Level
Data Link Control (HDLC), Frame Relay, and ATM. This lesson describes AToM
characteristics, implementation, and verification.

Objectives
Upon completing this lesson, you will be able to describe AToM. You will be able to meet
these objectives:
Introduce AToM
Implement AToM
Introduction to AToM
This topic explains AToM.

Subset ofVPWS:
- MPLStransport
- Point-to-point Layer 2 connections
Provisioning:
- Directed LOP requires unsummarized /32 PE loop back addresses
Forwarding:
- No MAC learning
- All ingress frames transported to the other end
Signaling:
- Setup, maintenance, and teardown of VCs and VC labels
- VCCV
- Directed LOP
MTU considerations:
- Fragmentation in core black-holes traffic
- Same MTU values on ingress and egress

AToM is a subset of Virtual VPWS that enables point-to-point Layer 2 virtual connections over
an MPLS infrastructure. Several aspects require special consideration:
Provisioning
Forwarding
Signaling
Maximum transmission unit (MTU) considerations
These aspects are discussed in this topic.

4-26 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 1. Use the xconnect command on ingress PE (port, subinterface, etc.).
2. PE1 starts a directed LOP session to PE2 (if not yet available):
- One LDP session can signal multiple PWs.
3. PE1 allocates the ve label and binds to the ve 10:
- SameVC ID on both ends; VC label unique per PE
4. PE1 sends mapping message (ve FEe TLV, ve label TLV).
5. PE2 receives ve FEe and label TLV and maps to local ve 10.
6. PE2 repeats the process (1 to 4, and then 5 on PE 1).

, Native service:
;:C ~ : + . - -Pseudo-wire
- - - - - - - - -..... ! . Native service.
!
'!'-eLI~~
:..
,' ,

RJf"~ CE

PE1 ~

AToM provisioning occurs in these steps:

1. The xconnect command is issued on the ingress provider edge (PE) (port, subinterface, and
so on). Alternatively, the pseudowire (PW) autodiscovery mechanism can be used to detect
the neighbor.

2. PEl starts a directed Label Distribution Protocol (LDP) session to PE2 if one is not yet
available. If a directed session already exists to the destination PE, it is reused for another
PW. One LDP session can signal multiple PWs.

3. PEl allocates a virtual circuit (Ve) label and binds it to the VC ID. The same VC ID value
must be configured on both ends. The VC label is unique per PE.

4. PEl sends a mapping message containing the VC forwarding equivalent class (FEe) type,
length, and value (TLV) and VC label TLV to the other end.

5. The other end (PE2) receives VC FEC and label TLV and maps it to the locally configured
VCID.

6. The other end (PE2) repeats the process (Steps 1-4), which is then finished on PEl by
receiving the VC FEC and label TLV and mapping it to the locally configured VC ID.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-27
 FEe
Set of packets handled in the same way on MPLS LSR
Used to bind a VC label to aVe 10
Multiplexing customer data over the same LSP tunnel

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

An Interior Gateway Protocol (IGP) is required in the MPLS backbone. All routers in the
backbone have routing information about how to send IP packets to each other. LDP is also
used between directly connected neighbors. Local labels are assigned to each IGP-derived
route. The label values are then propagated to the neighbor across the LDP session.
The IGP, together with the LDP sessions between directly connected neighbors, establishes
label-switched paths (LSPs) from any router inside the backbone to any other router inside the
backbone. In the figure, a unidirectional LSP from the upper right to the lower left is
established between the ingress and egress PE routers. The tunnel label is used to propagate the
packets along the LSP to the correct egress PE router.
The figure also shows the directed multihop LDP session that is used to exchange the VC label
between the ingress and egress PE routers. Any ingress-egress PE router pair will need such an
LDP session. In this example, the egress PE router allocates the label value 17. The VC label is
advertised to the ingress PE router using the directed LDP session between them.
The ingress PE router now forms a label stack. The topmost label, the tunnel label, has the
value 21 and is used to guide the packets to the egress PE router. The second label, the VC
label, has the value 17 and is used by the egress PE router to propagate out the packets on the
correct interface.
The ingress PE router receives a Frame Relay frame on data-link connection identifier (DLCI)
101 on the incoming interface. The DLCI is mapped to the AToM tunnel across the backbone.
The Frame Relay frame is therefore encapsulated into MPLS using the label stack with label 21
as the topmost label and label 17 as the second label.
The packet is then forwarded along the LSP. The topmost label is used for label swapping in
the next hop. The top label is changed to the value 22. In the next hop, label swapping results in
label value 23 being the top label. In the router just before the egress router, the incoming label
value 23 indicates pop. That label therefore performs penultimate hop popping (PHP). The
topmost label is removed, and the packet is propagated to the egress PE router with the label
value 17, the VC label, which is now the only label left.

4-28 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 When the PE router receives the packet with label value 17, that label value instructs the PE
router to decapsulate the packet and send it out on the associated Frame Relay DLCI. In this
case, the DLCI value is 202. The Frame Relay frame is now reconstructed and transmitted.

Establishing, maintaining, and tearing down VCs:

- Directed LDP signaling
- Frame Relay must use LMI procedures.
- ATM should use ILMI procedures.
If PE detects an event that affects service, it must withdraw VC label.

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

The signaling of the VC occurs over the targeted LDP sessions and includes the setup,
maintenance, and teardown of the PW. In addition, the PE routers translate the events from the
local attachment circuit to the PW and vice versa. This translation depends on the encapsulation
of the local attachment circuit.
A PE router may provide circuit status signaling on the interface where the customer connects.
A PE router that provides Frame Relay services must use Local Management Interface (LMI)
procedures with the customer equipment. A PE router that provides ATM services should use
Integrated Local Management Interface (ILMI) procedures.
An example of the signaling procedure, shown here, is the VC teardown. If a PE router detects
a condition that affects normal service, it must withdraw the corresponding VC label.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-29
 Layer 2 VPN OAM featu re
Keepal ive protocol to monitor PW data forwarding
AToM VCCV categories:
- Switching modes-for differentiating between control and data traffic
In-band (type 1) -uses PID field in the AToM control word to identify VCCV
control packet
Out-of-band (type 2) -MPLS router alert label is carried above the VC label
to identifyVCCV control packet
- Applications-in-band keepalive method
MPLS LSP ping
ICMP ping

Virtual Circuit Connectivity Verification (VCCV) is Layer 2 VPN Operation, Administration,

and Maintenance (OAM) feature that allows network operators to run a PE-to-PE keepalive
protocol across a specified PW to ensure that the PW data path forwarding does not contain any
faults.
When a PW is first signaled using LDP or L2TPv3, a message is sent from the initiating PE to
the receiving PE. That message has been extended to include VCCV capability information that
indicates to the receiving PE which combinations of control channel and connectivity
verification types it is capable of receiving. If the receiving PE agrees to establish the PW, it
will return its capabilities in the subsequent signaling message.
When MPLS is used to transport PW packets, VCCV packets are carried over the MPLS LSP.
Packets are sent across this channel either as in-band traffic with the data of the PW, or out-of-
band.
Two types ofVCCV switching modes have been defined to distinguish VCCV packets from
regular data traffic:
In-band VCCV (type 1): Control channel type 1 uses a protocol ID (PID) field in the
AToM control word to identify an AToM VCCV packet. This type ofVCCV is used for
those PW types that employ the control word.
Out-of-band VCCV (type 2): Control channel type 2 is also referred to as the MPLS
router alert label. The MPLS router alert label is carried above the VC label to identify an
AToM VCCV packet. Control channel type 2 can be used whether the PW is set up with a
control word present or not.

Cisco routers use type 1 switching, if available, when they send MPLS LSP ping packets over
an AToM VC control channel. Type 2 switching accommodates those VC types and
implementations that do not support or interpret the AToM control word.

4-30 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 An AToM VC advertises its AToM VCCV disposition capabilities in both directions-that is,
from the originating router (PEl) to the destination router (PE2), and from PE2 to PEL In some
instances, AToM VCs might use different switching types if the two endpoints have different
AToM VCCV capabilities. If PEl supports type I and type 2 AToM VCCV switching and PE2
supports only type 2 AToM VCCV switching, there are two consequences:
LSP ping packets sent from PEl to PE2 are encapsulated with type 2 switching.
LSP ping packets sent from PE2 to PEl use type I switching.

You can determine the AToM VCCV capabilities advertised to and received from the peer by
entering the show mpls 12transport binding command (Cisco lOS Software) or show 12vpn
xconnect all detail command (Cisco lOS XR Software).
Two verification types can be negotiated by the in-band or out-of-band VCCV control channel
in an MPLS environment:
ICMP ping: When this optional connectivity verification mode is used, an Internet Control
Message Protocol (ICMP) echo packet (ICMPv4 or ICMPv6) achieves connectivity
verification.
MPLS LSP ping: This method helps monitor LSPs and quickly isolate MPLS forwarding
problems.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-31
 AToM transport of Frame Relay, Ethernet, andAAL5 does not allow
packets to be fragmented and reassembled.
Ensure that the MTU of all intermediate links between endpoints is
sufficientto carry the largest Layer2 frame received.
The ingress and egress PE routers must have the same MTU value .

.......
Back-l.I' FR R Label (VC) EXP S TTL 4 bytes
/
.......
TE for FRR Label (VC) EXP S TTL 4 bytes
/
Core LOP Label (I.e) EXP S TTL
....... 4 bytes
/
.......
VC directec LOP Label (VC) EXP S TTL 4 bytes
/
.......
Optional controllMlrd 4 bytes
/
....... 4 bytes
Dot1 Q Header (only in Port Mode Xoonnect)
/
Ethernet POU
....... Upto1514
bytes
/

Unlike JP, most Layer 2 protocols (for example, Frame Relay, Ethernet, and ATM adaptation
layer 5 [AAL5]) do not allow fragmentation of frames. This fact has two implications:
All intermediate links between the ingress PE router and the egress PE router must be able
to carry the largest Layer 2 frame that has been received, including the imposed label stack
and the 4-byte control word (if it is used).
The ingress PE interface and the egress PE interface must have the same MTU value.

Failure to comply with the first rule means that the larger frames, where the label stack and the
control word contribute to creating a larger size than can be carried, will be dropped by the
backbone.
Failure to comply with the second rule means that frames that are forwarded along the LSP will
be dropped by the egress PE if the frame size is too large for the egress PE interface.
The figure illustrates the potential overhead imposed by various labels and headers. The MTU
size on the core must be able to accommodate the customer MTU size with the added overhead.

4-32 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Controlword is optional.
Transmitted after the label or labels and before the Layer 2 PDU
Can be used for in-band VCCV
Flag field carries different bits for different Layer 2 protocols:
- Frame Relay: FECN, BECN, DE, C/R
- ATM: AAL5 or cell, EFCI, CLP, C/R
SequencenumberO indicates that no sequencing is done.

Label (LDP) ExP 0 TTL Tunnel label

Label (VC) EXF 1 TTL VCllabel

Control word
0000
I Flags I Length
I Sequence Number
(Optional)

Layer2 PDU

The figure illustrates AToM encapsulation. The topmost label is the tunnel label. This label is
followed by the VC label. Following the VC label is the optional control word. Next in the
packet is the Layer 2 protocol data unit (PDU).
The control word is optional. Some Layer 2 protocols make use of it, while others do not. Both
endpoints (ingress PE and egress PE) must agree to either use or not use the control word. It is
transmitted after the label stack but before the Layer 2 PDD. It can be used to carry important
Layer 2 header information and to guarantee sequenced delivery, if required.
The control word is 32 bits long. It is divided into four fields. The first field is 4 reserved bits,
which must always be set to zero. The next field is a 4-bit flag field. The flags have different
uses depending on the Layer 2 protocol that is being forwarded. The third field is an 8-bit
length field, which is used only if the Layer 2 PDU is shorter than the minimum MPLS packet
and padding is required. If no padding is required, the length field is not used. The fourth field
is a 16-bit sequence number. Sequence numbering is used only on Layer 2 protocols, and it
guarantees ordered delivery. A special value of 0 in the sequence field indicates that there is no
guaranteed sequenced delivery.
When AToM is used for Frame Relay over MPLS (FRoMPLS), the Frame Relay header is
removed and the forward explicit congestion notification (FECN), backward explicit
congestion notification (BECN), discard eligible (DE), and command/response (C/R) bits are
carried in the control word flag field.
When AToM is used for ATM over MPLS, the first flag in the control word flag field is used to
indicate whether it is AAL5 frames or raw ATM cells that are being transported by AToM. The
other three flags are used for explicit forward congestion indication (EFCI), cell loss priority
(CLP), and C/R. If one ofthese flags is set in any of the ATM cells that are being transported in
the MPLS packet, then the corresponding flag is set in the control word.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-33
 ~-
o

o
...-
Layer 2 VPN extendsVCs over single service providerAS.
Changes in control and data plane code are required for inter-AS span.
o PW stitch ing solutio n:
- Interconnects PWs in different autonomous systems
- ASBRs are the stitch points
- Interworking of control and data planes at stitch point

(FR, Etrernet, Pseudo- (FR, Ethernet, ;

HDLC, PPP, ATM)_ Pseudowire wire i+._ps_e_Ud_O_W_ire_ _ i ,HDLC, PPP,AT~) :
- - - - 1 + - - - - -.. I'

CE
'---8 AS 65001 8-1
PE ASBR ASBR PE CE

Traditionally, AToM enables VC connections through a single autonomous system (AS). To

extend this function to interautonomous system (inter-AS) deployments, changes in the control
and data would be required. Pseudowire (PW) stitching provides an immediate solution.
PW stitching enables the extension of Layer 2 VPN PWs across an inter-AS boundary or across
two separate MPLS networks. Layer 2 VPN PW stitching connects two or more contiguous PW
segments to form an end-to-end multihop PW. This end-to-end PW functions as a single point-
to-point PW. Layer 2 VPN PW stitching enables the service provider to keep the IP addresses
of the edge PE routers private across inter-AS boundaries. Using the IP addresses of the
Autonomous System Boundary Routers (ASBRs), the ASBRs join the PWs of the two domains.
AToM packets forwarded between two PWs are treated the same as any other MPLS packet,
with the following exceptions:
The outgoing VC label replaces the incoming VC label in the packet. New IGP labels and
Layer 2 encapsulation are added.
The incoming VC label Time to Live (TTL) field is decremented by one and copied to the
outgoing VC label TTL field.
The incoming VC label experimental (EXP) value is copied to the outgoing VC label EXP
field.
The outgoing VC label "bottom-of-stack" S bit in the outgoing VC label is set to 1.
AToM control word processing is not performed at the Layer 2 VPN PW switching
aggregation point or ASBR. Sequence numbers are not validated.

4-34 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Ethemetframes are transported without preamble, SFD, and FCS.
In Ethemet port mode, all VLAN information is transmitted:
- May be overwritten by the egress PE
Controlword is optional.

Ethernet II Encapsulation
<7 octets> <1 octet> <6 octets> <6 octets> <2octets> <2octets> <2octets> <46-1500> <4octets>

DA SA EtherTwe Data

~ ~
~ T~",,,,rt'" ",",AToM ~

DA SA Lergth Data

<7octets> <1octet> <6actets> <6actets> <2octets> <2octets> <2octets> <3octets> <3octets> <2octets> <46-1492> <4octets>

802.3/802.2/SNAP Encapsulation

AToM encapsulates entire frames but excludes fields whose transmission does not offer any
benefits, such as frame synchronization data. In Ethernet over MPLS (EoMPLS), the preamble,
the start frame delimiter (SFD) and the frame check sequence (FCS) are excluded from
encapsulation.
The preamble of an Ethernet frame consists of a 56-bit (7-byte) pattern of alternating 1 and 0
bits, which allows devices on the network to easily detect a new incoming frame. The SFD is
designed to break this pattern, and signal the start of the actual frame. The SFD is the 8-bit (1-
byte) value marking the end of the preamble of an Ethernet frame. The SFD is immediately
followed by the destination MAC address. It has the value 10101011. The FCS provides Layer
2 checksum characters added to a frame for error detection and correction purposes.
In Ethernet port mode, all VLAN information is transmitted. The VLAN tag or tag stack may
be overwritten by the egress PE. The egress PE can also manipulate the VLAN tag received
over the vc. The control word is optional in EoMPLS.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-35
 Failures 1 and 2 (transit network):
- IGP and MPLS LDP will reconverge.
- With MPLS traffic engineering and FRR enabled, failover to backup tunnel.
- PW will stay up as long as PE1 has available LSP path to PE2.
- PW service layer is not affected.
Failures 3 and 4 (service node or attachment circuit):
- EoMPLS PW will go down.
- Network transport layer reconverge does not help.
Solution: PW redundancy Prin;ary PW Attachment
/
/ PE2 / /Circuit

" "

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

It is common practice to add redundancy to the interconnect between two customer sites to
avoid split-subnet scenarios and service interruption. The redundancy solution comprises
several building blocks. One block refers to the redundant LAN access and considers issues
related to the Spanning Tree Protocol (STP). This component is described in the next lesson.
This figure presents a redundant transit network and redundant attachment circuits. In case of
failure 1 or 2, as shown above, IGP and MPLS LDP will reconverge. MPLS traffic engineering
and Fast Reroute (FRR) will trigger an automatic failover to the backup tunnel. The PW will
stay up as long as PEl has an available LSP path to PE2. This failure scenario does not affect
the PW service layer.
In case of failure 3 or 4, as shown above, the PW will go down. Network layer reconvergence
will not help re-establish the service. A solution has been developed to address this problem. It
is referred to as PW redundancy.

4-36 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 --
Dual-homing of one local PE to:
- Two remote PEs
- Two different attachment circuits on the same remote PE
Two PWs: Primary and backup provide redundancy for a single
attachment circuit or node.
Faults on the primary PW cause failover to backup PW.

Case 1: SelVice node and Case 2:

attachment circuit Attachment circuit
protection protection

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

PW redundancy can be implemented using a one-way or two-way method. In one-way

redundancy, the local PE has two PWs to two remote PEs serving the same destination site or
to the same egress PE with redundant attachment circuits. One PW is declared as primary, the
other as backup. A fault of the primary PW triggers a fallover to the backup PW.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-37
 o

o
--
Dual-homing of two local PEs to two remote PEs
Four PWs:
- One primary PW
- Three backup PWs
o Requires Me-LAG
- Point of attachment nodes run ICCP
- ICCP synchronizes state and forms a redundancy group.

Active PW
Active PoA I Active PoA

................
I
~

leep
....... . .
.
00

;'IE:"iII~
................
..
......- " - - I

......r-_......
Standby PoA I ~-- St db P A
Standby PWs (3) an y 0

C2012Ciscoamlloritsafflll_.AII~ghtll"",erved.

In two-way redundancy, four PWs are used to provide high availability service. Only one PW is
declared as primary. The three remaining PWs are intended for backup. In each site, one
attachment circuit is primary, the other backup. To synchronize the redundancy information
between the LAN and the PE devices, Multi-Chassis Link Aggregation Group (MC-LAG) must
be enabled in the network. MC-LAG refers to the PE devices as points of attachment nodes.
The points of attachment run Inter-Chassis Communication Protocol (lCCP) to synchronize
state and form a redundancy group.

4-38 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Service instances configured on main interface:
- Also known as EFP
Each EFP matches a predefined VLAN tag-based criteria.
Optional tag manipulation can be configured.
Traffic forwarding is specified.
Features such as QoS policies can be specified.

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

Ethernet Virtual Connection (EVe) is used to represent Cisco software architecture to address
Carrier Ethernet services.
Cisco implements EVC using Ethernet Flow Points (EFPs). An EFP is a substream partition of
a main interface. On Cisco routers, the EFP is implemented as a Layer 2 subinterface with an
encapsulation statement.
The EVC solution defines these aspects of Ethernet-based attachment circuits and virtual
circuits:
Frame matching based on one or more VLAN tags
Optional VLAN tag manipulation
Traffic forwarding
Additional services, such as quality of service (QoS) policies

EVC is not supported on all Cisco platforms. It is supported on Cisco lOS and lOS XR
Software.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-39
 Multiple Layer 2 frame types - - - Multiple Layer 2 services

Untagged
Single-tagged
Customer Double-tagged
Network 802.1q
802.1ad

Access side: Trunk side:

Customer Ethernet attachment - Local Layer 2 cross-connect
circuit - Local Layer 2 bridging
- Terminates on an EFP - EoMPLS or VPWS
- VPLS or H-VPLS
- Layer 3 routing
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Customer-to-service mapping is illustrated here as a connection between an Ethernet-based data

flow on the customer side and a service on the trunk side.
Flexible Ethernet mapping is the ability to process and classify different Ethernet frame types,
each with different attributes (EtherTypes, VLAN tags, class of service [CoS] bits, and so on).
Cisco lOS XR Software uses the Ethernet Flow Point (EFP) concept to provide flexible
Ethernet mapping. Flexible transport is found on the trunk side. Each Ethernet flow from the
customer side is mapped or connected to a service on the trunk side. These service types can be
native Ethernet, IP, or IP and MPLS-based, and they form the basis for Layer 2 VPN.

4-40 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 EFPs enable flexible
mapping of frames into InnerVLAN
OuterVLAN
Layer 2 services. tag tag
Mapping is based on VLAN
tagging:
802.1 Q, 802.1 ad
Single-tag or double-tag
Unique or multiple values
(ranges or lists)
Untagged traffic
Unclassified traffic (default)

iCl2012Ciscoamlloritsatrlllllte!l.A1I~ghtll"",erved.

One of the EVe advantages is flexible frame matching. Flexible frame matching is a
functionality that allows each service instance to match frames with either a unique single
VLAN, or a list or range ofVLANs. It can also match single- or double-tagged frames,
untagged frames, or any frames that are not matched by the specific statement.
Flexible frame matching is the first step when configuring a service instance. Subsequent steps,
encapsulation rewrite, and forwarding definition are described in the following pages.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-41
 Eve supports only nonexact matching.
encapsulation dot1 q 10 matches any packets with outmost tag
equal to 10:

encapsulation dot1 q 10 second 100 matches any packets with

outmost tag equal to 10 and second outmost tag equal to 100:

10 100 1----:.1..:.,.00:...,:0----l.1 _

Several subinterfaces with various matching statements can exist on a main interface. The
matching logic affects the assignment of a frame to a specific subinterface. The Eve supports
only nonexact matching, in which additional inner-VLAN tags are not taken into consideration.
The command encapsulation dotlq 10 matches any frames with the outmost VLAN tag equal
to 10. This includes both 801.1q frames, queue-in-queue (QinQ) frames with any second
VLAN tag, as well as frames with more than two VLAN tags, if the outmost tag is 10.
The command encapsulation dotlq 10 second 100 matches any frames with the outmost
VLAN tag equal to 10 and second outmost tag equal to 100. This includes frames that contain
more than two VLAN tags, if the first two tags fulfill the defined criteria.

4-42 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Longest match defines frame-to-EFP matching.

~L-- _ _-------.J
\

I 10 I 200 ---- ) dot1q 10

10 100

1
Frame received EFP configuration

iCl2012Ciscoamlloritsafflllllte!l.A1I~ghtll"",erved.

If several potential matches exist on the same main interfaces, the longest-match rule defines
the most specific hit.
With the three subinterfaces shown above, the first one is the only subinterface that matches the
first two frames (single tag 10, and double tag 10-200).
The second frame, with double tag (10-100) is matched by the first and second subinterface.
Because the second subinterface defines a more exact match, it will process the second frame.
The third frame, with double tag (10-130) is matched by the first and third subinterface.
Because the third subinterface defines a longer match, it will process the third frame.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-43
 Eve provides several VLAN tag rewrite options.
Push:
- Adds one or two VLANs to traffic
- push {dot1q <vlan-id> I dot1q <vlan-id> second-dot1q <vlan-id>}
Pop:
- Removes one or two VLANs from frames
- pop {112}
Translate:
- 1-to-1 dot1q <vlan-id>
- 2-to -1 dot1 q <via n-id>
- 1-to-2 dot1q <vlan-id> second-dot1q <vlan-id>
- 2-to-2 dot1q <vlan-id> second-dot1q <vlan-id>
Symmetric keyword allows simplicity and avoids misconfiguration.

Once a packet is matched and assigned to a given subinterface, its VLAN tags can be modified.
You can configure several of these three operations:
Push: This action adds one or two VLANs to traffic using the push {dotlq <vlan-id> I
dotlq <vlan-id> second-dotlq <vlan-id>} command.
Pop: This action removes one or two VLANs from frames.
Translate: This action allows the replacement of one or two tags by another one or two
tags, using one ofthese commands:
1-to-1 dotlq <vlan-id>
2-to-1 dotlq <vlan-id>
1-to-2 dotlq <vlan-id> second-dotlq <vlan-id>
2-to-2 dotlq <vlan-id> second-dotlq <vlan-id>
The symmetric keyword defines that the pop, push, or translate operation is reversed for egress
frames.

4-44 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Three forwarding options through EFP:
Local con nect
- Point-to-point connections between two EFPs on same router
Scalable EoMPLS
- Point-to-pointXconnect between two EFPs on different routers
Bridge domain
- Classical Layer 2 switching domain
- Can be integrated with VP LS or Layer 3 IP address (IRS)
- Split horizon can be configured on the bridge domain.
EFP and subinterfaces can coexist on the same physical interface.

Lastly, the Eve framework is used to define traffic forwarding. Incoming frames can be
transmitted to these destination types:
Local connect: This destination represents a point-to-point connection between two EFPs
on the same router. No VLANs are required to be defined on the PE device.
EoMPLS: This destination refers to a point-to-point Xconnect between two EFPs on
different PE routers. No VLANs are necessary on the PE device.
Bridge domain: This destination describes a Virtual Private LAN Service (VPLS) that
uses a classical Layer 2 switching domain. The associated switch virtual interface (SVI)
can be enabled for a VPLS or Layer 3 IP address (integrated routing and bridging [IRBJ).
You can enable or disable the split-horizon rule within the bridge domain.
EFP and subinterfaces can coexist on the same physical interface.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-45
 3
; - - - - - - ..... \
VLAN tag local
significant

Flexible VLAN
tag classification
Flexible VLAN
tag rewrite
Flexible
EtherType (.1Q,
QinQ, .1ad)
I
'- -0 - -.;

Layer 2 or Layer 3 ~- - - - - - - - - - - - - - - - - - - - - - - - - - - - "

subinterfaces ~ Flexible service mapping and multiplexing. Support all standard-based I
(802.1a/QinQ/.1ad) U' services concurrently on the same port I
I Layer 2 peer-to-peer local connect and EoMPLS I
I Layer 2 multipoint local bridging, H-VPLS and VPLS I
" _ ~(Lul~r .!:a~e.!:. 3_s!!.b!!:lt~rf~c~ ~n!! i~te.9r~t~d .!:a~e.!:. 2_a~dl-~! ~-.!R. /
C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

This diagram summarizes the Eve functionality by showing a physical Gigabit Ethernet
interface with multiple EFP-based subinterfaces. The subinterfaces can be independently
configured with flexible VLAN matching and VLAN tag rewriting. The incoming frames are
then forwarded according to the defined parameters: linked to local attachment circuits, bridged
or switched over a PW, or routed.

4-46 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
AToM Implementation
This topic describes how to implement AToM.

Prepare MPLS infrastructure:

- P E routers must have a /32 address on their loop backs.
- PE loopback addresses cannot be summarized in the core.
- MPLS enabled in the core (unless L2TPv3 is used).
- Ensure MTU sizes in the core are large enough.
Enable Layer2 frame transport on both endpoint PE routers.
Make sure MTU is the sameon both endpointinterfaces.
Optionally configure parameters:
- Port or VLAN mode, control word, sequencing, and so on
Optionally configure AToM interworking.

To implement AToM, you first need to prepare the MPLS infrastructure. This includes these
steps:
PE routers must have /32 addresses on their loopbacks. This is required for binding the VC
label and the transport LDP into a label stack.
The PE loopback addresses cannot be summarized in the core. Aggregation in the network
would break the LSP path.
MPLS must be enabled in the core unless L2TPv3 is used to provide IP-based transport.
MTU sizes on the core links must be able to accommodate the customer MTU with the
added label overhead.

Once the MPLS infrastructure is ready, you will do the following:

Enable Layer 2 frame transport on both endpoint PE routers.
Make sure MTU is the same on both endpoint interfaces.
Optionally configure parameters, such as port or VLAN mode, the control word,
sequencing, and so on.
Optionally configure AToM interworking.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-47
 105 XR
CE1 CE2

Cisco lOS XR:

Cisco lOS and lOS XE:
ip cef
mpls ip
mpls label protocol Idp
interface LoopbackO
mpls Idp router-id LoopbackO force
ipv4 address 10.1.1.1 255.255.255.255
interface LoopbackO
interface GigaOjOjOjO .40 12transport
ip address 10.2.2.2 255.255.255.255
encapsulation dotlq 40
!
12vpn pseudowire-class pw-class2
encapsulation mpls
xconnec t group eompls -group
p2p eompls-p2p
interface GigabitOjOjOjO.40 interface GiG/O/O .40
encapsulat ion do tlQ 40
neighbor 10.2.2.2 pw-id 123
xconnect 10.1.1.1 123 pw-class pw-class2

ControlllJOrd is optional in EoM PLS

EoMPLS is the most common AToM flavor. The configuration on Cisco lOS XR platforms
differs from the one on Cisco lOS or lOS XE platforms.
On both systems, the pseudowire (PW) class is used as a container for optional parameters,
such as encapsulation type, the use of control word, transport mode (port or VLAN),
sequencing, and others.
On Cisco lOS XR Software, the attachment circuit must be enabled for Layer 2 transport using
the 12transport keyword. The PW is configured using the xconnect group and p2p commands
in circuit configuration mode. The neighbor command specifies the remote PE address and the
VCID.
On Cisco lOS and lOS XE platforms, the xconnect command in interface configuration mode
defines the remote end, the VC ID, and, optionally, the PW class. On lOS XR platforms, the
xconnect command is found in Layer 2 VPN configuration mode.
To specify a preferred interface for determining the LDP router ID, use the mpls Idp router-id
command in global configuration mode.
The normal (default) method for determining the LDP router ID may result in a router ID that is
not usable in certain situations. For example, an IP address that is selected as the LDP router ID
might be an address that the routing protocol is not able to advertise to a neighboring router.
The mpls Idp router-id command provides a means for specifying an interface whose IP
address is to be used as the LDP router ID. The specified interface must be operational for its IP
address to be used as the LDP router ID.
When it is executed without the force keyword, the mpls Idp router-id command modifies the
method for determining the LDP router ID by causing the selection of the IP address of the
specified interface argument (if the interface is operational) the next time that it is necessary to
select an LDP router ID. The effect of the command is thus delayed until this need arises,
which is typically the next time that the interface whose address is the current LDP router ID is
shut down or the address itself is not configured.

4-48 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 When it is executed with the force keyword, the effect of the mpls ldp router-id command
depends on the current state of the specified interface:
If the interface is up (operational) when the mpls ldp router-id force command is issued
and if its IP address is not currently the LDP router ID, the LDP router ID is forcibly
changed to the IP address of the interface. This forced change in the LDP router ID tears
down any existing LDP sessions, releases label bindings that have been learned via the
LDP sessions, and interrupts MPLS forwarding activity that is associated with the bindings.
If the interface is down (not operational) when the mpls ldp router-id force command is
issued, when the interface transitions to up, the LDP router ID is forcibly changed to the IP
address of the interface. This forced change in the LDP router ID tears down any existing
LDP sessions, releases label bindings that have been learned via the LDP sessions, and
interrupts MPLS forwarding activity that is associated with the bindings

Note The VC 10, control word usage, and MTU sizes and PW type must match on the interfaces
at both ends.

CE1 PE1 PE2 CE2

.I--~~

pseudowire-class pw-classl p seudowire -cl ass pw-c lass2

encapsulation mpls encapsulation mpls
con tro I-word con tro I-word

interface LoopbackO interface LoopbackO

ip address 10.1.1.1 255.255.255.255 ip address 10.2.2.2255.255.255.255

interface seriall/O interface ser iall/O

no ip addres s no ip addres s
encapsulation pppJhdlc encapsulation pppJhdlc
xconnect 10.2.2.2 123 pw-class pw-classl xconnect 10.1.1.1 123 pw-class pw-class2

Controlword is optional in PPP and HDLC over MPLS

PPP and HDLC over MPLS is configured using the same commands as EoMPLS. This scenario
shows Cisco lOS and lOS XE routers at both ends, because serial interfaces are more common
for this router range.
The use of the control word is optional in PPP and HDLC over MPLS.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-49
 The attachment circuits are terminated locally.
There are two types of interworking (any-to-any):
- Ethernet (bridged):
Ethernetframes are extracted from attachment circuit and sent over PW.
VLAN tag is removed.
CEs can run Ethernet, BVI, or RBE.
Use the interworking ip command.
- IP (routed):
IP packets are extracted from attachment circu it and sent over the PW.
Use the interworking ethernet command.
.-.'I!S'!llffl"&.''lIf~j!'!!i~ti~'.l'P!li:a~t"&'
."....i("i.I;,i4.
Frame Relay to EthernetNLAN Yes Yes Yes Yes

Frame Relay to PPP Yes Yes Yes No

Frame Relay to ATMAAL5 Yes No Yes No

EthernetNLAN to ATM AAL5 Yes No Yes Yes

Ethernet to VLAN Yes Yes Yes Yes

AToM interworking links two disparate Layer 2 encapsulations by terminating each attachment
circuit locally and binding them over the PW.
Interworking can be implemented using bridging or routing.
In bridging mode, the frames are extracted from the attachment circuit and sent over the PW. In
the case of Ethernet, the VLAN tag (if available) is removed. In that case, the CEs can run
Ethernet, bridge-group virtual interface (BVI), or routed bridge encapsulation (RBE).
In routing mode, the IP packets are extracted from the attachment circuit and sent over the PW
to the remote end.
The figure lists various interworking methods and their feasibility in AToM, L2TPv3, routing,
and bridging mode.

4-50 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 ~._~---
CE1 PE1 PE2 CE2

PE1:
frame-relay Bwitching PE2:

pseudowire-clllBB lltom_fr_vlan pBeudowire-clllBB IItom_vlan_fr

encapsulation :lip Is encapsuillti on mpls
interworking ip interworking ip

interface serioal3/0 interface GigabitBthernet4jO. 310

encapsulation frame-relay encapsuillti on dotlQ 310
clock source int enlal xconnect 10.1.2.1210 pw-class atom_vlan_fr
frame-relay Imi- type ansi
frame-relay intf -type dee
!
connect fr-vlan s eria13/0 210 12 transport
xconnect 10.1.2.2 210 pw-class atom_fr_vlan

CE1: CE2:
interface serialS /0. 210 point-to -point interface GigabitBthernet6/0.310
ip address 172.16.1.1 255.255.255.0 encaps ula tion dotlQ 310
frame-relay interfllce-dlci 210 ip address 172.16.1.2 255.255.255.0

These configurations depict a Frame Relay-to-Ethernet interworking scenario using routing

mode. The mode is defined with the interworking ip command in the pseudowire-class
configuration mode.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-51
 f!l Cisco lOS XR:

RPjOjRSPOjCPUO:PEl# show 12vpn xconnect

Legend: ST = State, UP = Up, DN" = Down, AD = Admin Down, UR = Unresolved,
LU = Local Up, RU = Remote Up, CO = Connected
XConnect Segment 1 Segment 2
Group Name ST Description ST Description ST

eompls -group eompls-p2p UP Gigabi to /OjO/O .30 UP 10.2.2.2 123 UP

Cisco lOS and lOS XE:

PE2#show xconnect all detail

Legend: XC ST=Xconnect State Sl=Segmentl State S2=Segment2 State
UP=Up DN=Down .AD =Admi n Down IA=Inactive
SB=Standby HS=Hot Standby RV=Recovering NH=N"o Hardware

XC ST Segment 1 81 Segment 2 S2
--- --- +-- --- --- -- --- -- -- --- --- --- -- -- --- +-- +-- --- - --- --- --- -- --- -- -- --- --- --- +--
UP ac GiOjOjO.40,40(Eth VLAN) UP mpls '0.1.1.,,'23 UP
Interworking: none Local VC label 16003
Remote VC label 30005
pw-class: pw-class2

The EoMPLS operation is verified using different commands on Cisco lOS XR and Cisco lOS
and lOS XE platforms. The show 12vpn xconnect command provides brief information on
configured cross-connects.
On Cisco lOS routers, you can display equivalent information with the show xconnect all
detail command.

4-52 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 RPjOjRSPOjCPUO:router# show 12vpn xconnect detail
Group eampls-group, xc eompls-p2p, state is up; Interworking none
AC: GigabitOjOjOjO.30, state is up
Type VLAN
MTU 1500; XC ID OxS 00 000 1; interworking none; MSTi 0
Sta tis tic s:
packet totals: send 90
byte totals: send 19056
PW: neighbor 10.2.2.2, PW ID 123, state is up ( established
PW class pw-classl, xc ID OxSOOOOOl
Encapsulation MPLS, protocol LDP
PW t'YPe VLAN, control word enabled, interworking none
PW backup disable delay a sec
Sequencing not set
MPLS Local Remote

Label 30005 16003 Cisco lOS XR

Group ID OxSO 00 300 Ox50 004 00
Interface GigabitOjOjOjO.30 GiO/O/O .40
MTU 1500 1500
Control word enabled enabled
PW t'YPe VLAN VLAN
vccv CV type Ox2 Ox2
(LSP ping verification) (LSP ping verification)
vccv CC type Ox 7 Ox7
(control word) control word)
(router alert label) (router alert label)

<output truncated>

Cisco lOS XR devices offer a much more granular output when you use the show 12vpn
xconnect command. The output includes information about the local (configured) and remote
(signaled) parameters, such as labels, group ID, attachment circuit interface, MTU size, control
word usage, PW type (port or VLAN), and VCCV connectivity verification and control channel
types. If any of these parameters do not match on both sides, the PW does not reach the up
state.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-53
 Cisco lOS and lOS XE:

PE2# show mpls 12transport vc detail

Local interface: GiOjOjO.40 up, line protocol up
Destination address: 10.1.1.1, VC ID: 123, VC status: up
Tunnel label: imp-null, next hop point2point
Output interface: POD/I/O, imposed label stack {16}
Create time: 00:16:44, last status change time: 00:15:45
Signaling protocol: LDP, peer 10.1.1.1:0 up
MPLS VC labels: local 16003, remote 30005
Group ID: local 12, remote 1
MTU: local 1500, remote 1500
Remote interface description:
Sequencing: receive disabled, send disabled
vc statistics:
packet totals: receive 56, send 55
byte totals: receive 10181, send 10569
packet drops: receive 0, send a

In Cisco lOS Software, you can investigate the VC-related information by issuing the show
mpls 12transport vc detail command. It displays the local and remote interfaces, exchanged
labels, group ID, MTU sizes, sequencing status, and VC statistics.

4-54 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Summary
This topic summarizes the primary points that were discussed in this lesson.

o EoMPLS is the most common AToM method that supports a host of

features, such as inter-AS operation, redundancy, and EVe
infrastructu reo
o AToM can be implemented in like-to-like fashion, or in any-to-any by
using AToM interworking.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-55
4-56 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Lesson 31

Implementing VPLS
Overview
Virtual Private LAN Services (VPLS) can be used to implement Carrier Ethernet in a service
provider MPLS infrastructure.
This lesson describes the VPLS implementation on Cisco lOS XR routers.

Objectives
Upon completing this lesson, you will be able to describe Ethernet services that are used in the
service provider network. You will be able to meet these objectives:
Discuss VPLS
Implement VPLS and H-VPLS
VPLS Overview
This topic discusses the VPLS implementation technique.

o Initial traffic across all PWs; MAC address is learned.

o Split-horizon forwarding is applied to avoid loops between PEs.
o Traffic is sentto relevant PWs (all or one).
o On PE failure, PWs go down and MACs are flushed.
o MAC learning process beginsagain.

HostS

Host A

Hoste

Forwarding of Unicast Frames

Suppose that the MAC address of host C is C and the MAC address of host B is B, for the
customer network in the figure. Suppose host C sends a frame with source MAC address C and
destination MAC address B. Suppose that PE3 does not know the location of MAC address B.
As a learning bridge would do, PE3 floods the packet on all ports except the port on which it
arrived. This means the packet is flooded to the pseudowire (PW) to PE2 and the PW to PEl.
PEl and PE2 know that the packet belongs to the customer VPLS, by virtue of the PW on
which the frame arrived. PEl and PE2 both perform destination MAC address lookup in their
VPLS forwarding tables corresponding to this customer. If PEl does not know the location of
MAC address B, it floods the frame on its local ports to the customer edge (CE). However, it
does not flood the frame to any other provider edge (PE) routers. This split-horizon scheme
ensures no forwarding loops occur. Similarly, PE2 forwards the frame to the port facing the
switch CEo
Receiving frames with MAC address C enables each PE to learn the location of host C. Thus,
PE2 and PEl create an entry in their forwarding tables with an association between MAC
address C and their respective PWs to PE3. In this way, all PEs learn the MAC addresses and
create an association between MAC addresses and PWs (for remote destinations) in their
forwarding tables for that particular VPLS instance.

Forwarding of Broadcast and Multicast Frames

Suppose PE3 receives a broadcast frame sent by host C. The frame must be sent to all sites of
the customer VPLS. PE3 floods the frame on PWs to PEl and PE2. In turn, PEl and PE2 flood
the frame to the attached CEs, but due to split horizon, do not send the frames to any PEs.
Multicast traffic is also treated this way.

4-58 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 .... --.-..=-..-
Each PE has a point-to-multipoint view of all other PEs:
- Sees itself as a root bridge with split-horizon loop protection
Full mesh topology obviates STP in the service provider network.
Customer STP is transparentto the service provider:
- CustomerBPDUs are forwarded transparently.

-:------l,I:,-I-.-------
___________ ~\ b?j /',"_ _ .
\c::
- .........JI~::I::tL_."
Ful mesh LOP
Ethernet PW to each peer

PEview

The full mesh ofPWs in the service provider network allows the implementation of the split-
horizon principle that is similar to the split horizon of an internal BGP mesh. Frames received
over a PW are generally not forwarded to another PW. This concept provides loop-free
forwarding. Therefore, there is no need for Spanning Tree Protocol (STP) in the service
provider network. The customer STP bridge protocol data units (BPDUs) are transparently
tunneled over the PWs to detect loops on the customer layer.

Note The split-horizon rule must be disabled in certain environments, such as hierarchical VPLS
(H-VPLS). H-VPLS is described later in the lesson.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-59
 Software featu re for:
Flooding and forwarding
- MAC table instances per customer (port or VLAN) for each PE
- Learning and forwarding process
- Flood unknowns, multi casts, and broadcasts to all other ports
Address learning and aging
- LOP enhanced with additional MAC list TLV (label withdrawal)
- MAC timers refreshed with incoming frames
Loop prevention
- Create full-mesh of PW VCs (EoM PLS)
- Split-horizon concept
- CustomerSTP BPOUs tunneled through the service providercloud
Implemented as VFI
- B ridge that connects attachment circuits to PWs
- VLAN extension

The Virtual Switch Interface (VSI), implemented as a virtual forwarding instance (VFI), is a
virtual Layer 2 Forwarding entity that defines the VPLS domain membership and resembles
virtual switches on PE routers. A VPLS domain consists of Ethernet interfaces or VLANs that
belong to the same (virtual) LAN but are connected to multiple PE devices. The VSI learns
remote MAC addresses and is responsible for proper forwarding of the customer traffic to the
appropriate end nodes. It is also responsible for guaranteeing that each VPLS domain is loop-
free. The VSI is responsible for several functions, namely MAC address management, dynamic
learning of MAC addresses on physical ports and virtual circuits (VCs), aging of MAC
addresses, MAC address withdrawal, flooding, and data forwarding.
All PWs in a VFI are placed by default into the same split-horizon group, which effectively
prevents traffic from forwarding to other PWs in the same VFI.
With VPLS Layer 2 VPNs, the customer can connect with a switch or a router.
If connecting with a router, the VPLS just looks like a switch to the routing protocols on each
side. No MAC learning is done beyond the MAC address of the directly connected CE router
interface.
If connecting at Layer 2 with a switch, then MAC learning on the PE of all the MAC addresses
on the directly connected customer LAN is possible. This is a situation where MAC limiting or
another filtering method would be implemented, or a provider backbone bridge (PBB)
(802.lah) can be used.

4-60 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 Nonhierarchical
Two architectures: MPLS Gore

. . . . .---.eJI;;$:
N-F1 :.~
...
- Nonhierarchical
Single PE (flat)
..::>.<:., VPLS

"~--:,~.:: ,
~
.. .. :::~
'
".~
- Hierarchical (H-VPLS)
With Ethernet access
- 802.1 ad (IEEE standard for QinQ) Hierarchical MPLS Gore

With MPLS access CE1

N-PE1 ~
Two PE roles: S:><:~
802.1 ad :"/,.: VPLS
- Network-facing PE (N-PE)
VPLS termination
~~L_.JlIIiiJ'~..::II""N_~PErt/::.:~'.::.
Layer 3 services
User-facing PE (U-PE)
Customer UNI

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Hierarchical VPLS (H-VPLS) topologies have been developed to address the scaling
limitations within flat VPLS architecture. VPLS requires a full mesh of tunnel LSPs between
all PE routers that participate in the VPLS service. For each VPLS service, n*(n-l)/2 PWs must
be set up between PEs. This creates signaling overhead. The packet replication requirement
deters large-scale deployment. Hierarchical connectivity reduces signaling and replication
overhead.
There are two access mechanisms in H-VPLS: Ethernet and MPLS. Ethernet-based access uses
the IEEE 802.lad standard or its predecessor, 802. 1QinQ.
H-VPLS defines user-facing PE (UPE) routers and network-facing PE (NPE) routers.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-61
 Significant signaling overhead Minimizes signaling overhead
Full PW mesh from the edge Full PW mesh among core
devices
Node discovery and
provisioning extends end to end Partitions node discovery
process
VPLS H-VPLS

PE CE
~ ~
NPE ~....----.
"PE ' . PE

. .~t~
/ :........... CE
CE CE

CE

ri(~,>:,~lil "'0"

CE
!!J!.t'
~
PE

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The H-VPLS hierarchy provides the benefits of less signaling in an MPLS core network and
less packet replication on NPE routers. The UPE routers have an aggregation role and do some
packet replication and MAC address learning.

4-62 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 802.1 ad is the IEEE standard for OinO.
802.1 ad outer EtherType: Ox88a8
802.10 Ethertype: Ox8100

Full mesh of
802.3 802.10 802.1ad pseudowires
:c----->i)~ ..
, -----'-->'~~c---------_) ~ c )

1C12012Ciscoandioritsafflllllte!l.A1I~ghts",served.

IEEE 802.1 QinQ is an Ethernet networking standard formally known as IEEE 802.1 ad. It is
also known as provider bridging, stacked VLANs, or queue-in-queue (QinQ).
802.1 QinQ specifies architecture and bridge protocols to provide separate instances of the
MAC services to multiple independent users in a manner that does not require cooperation
among the users, and requires a minimum of cooperation between the users and the provider.
The idea is to provide the possibility for customers to run their own VLANs inside the VLAN
of a service provider. This way, the service provider can just configure one VLAN for the
customer, and the customer can then treat that VLAN like a trunk.
IEEE 802.1 QinQ has these characteristics:
802.1 Q has a 12-bit VLAN ID field, which has a theoretical limit of 2'" 12=4096 tags. With
the growth of networks, this limitation has become more acute. A double-tagged frame has
a theoretical limitation of 4096*4096=16, 777,216, sufficient to accommodate network
growth.
The addition of a second tag allows new set of operations. Having multiple tags-the tag
stack-allows switches to more easily modify frames. In a tag stack scheme, switches can
add ("push"), remove ("pop") or modify single or multiple tags.
A multitagged frame not only has multiple VLAN IDs, but has multiple EtherTypes and
other VLAN header bit fields.
A tag stack creates a mechanism for Internet service providers to encapsulate customer
single-tagged 802.1 Q traffic with a single tag, the final frame being a QinQ frame. The
outer tag is used to identify and segregate traffic from different customers; the inner tag is
preserved from the original frame.
802.1 QinQ is upward compatible with 802.1 Q. Although 802.1 QinQ is limited to two tags,
there is no ceiling on the standard limiting a single frame to more than two tags, allowing
for growth in the protocol. In practice, service provider topologies often anticipate and
utilize frames with more than two tags.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-63
 1-," &
PW full mesh in core:
- Split-horizon for loop avoidance
Hub and spoke access PW for access:
- Only one PW per UPE (per service instance) active at
a time

Full mesh of
~;.-
, ----,8...:.;02:.;..3,--...;,: .,----'8...:.;02::.;...;.,.10:.:....-".;.,:,....;S.;.,.in;.:<.gl...:.;e...:.;or....;.r.;.,.ed...:.;u...:.;nd...:.;a...:.;nt-"-p.;.,.se...:.;ud,;;,;o....;-w,;;,;ir...:.;es"--+.: ' pseudowires,

....... ~:~~~~?;.~~.~~"Xim
_____
MPLS

Customer UPE
M~~:~9~ _sa in core

CEs
switches
Inactive pseudowire -.. ~
NPEs
One or several redundant
pseudowires to NPE

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

The second approach uses MPLS in the edge network, instead of 802.1 ad, which was described
in the previous method.
In the MPLS-based edge, the VPLS core PWs (hub) are augmented with access PWs (spoke) to
form a two-tier H-VPLS. The customer sites are connected to UPE devices. The UPE devices
have single or redundant connection (PW) to NPE routers. The NPE routers are connected in a
basic VPLS full mesh service. For each VPLS service, a single spoke PW is set up between
UPE and NPE devices. Unlike traditional PWs that terminate on a physical or logical port, a
spoke PW terminates on a VSI on UPE and NPE devices. The UPEs and NPEs treat each spoke
connection like an attachment circuit of the VPLS service. The PW label is used to associate
the traffic from the spoke PW to a VPLS instance.

UPE Operation
The UPE device supports Layer 2 switching and does all the normal bridging functions of
learning and replication on all its ports, including the spoke PW. Packets to an unknown
destination are replicated to all ports in the service, including the spoke PW. Once the MAC
addresses of CE devices connected to the same UPE device are learned, traffic between them is
switched locally, saving the capacity of the spoke PW to NPEs. Similarly, traffic between
remotely connected CE devices to different UPE devices is switched directly onto the spoke
PW and sent to NPEs over the point-to-point PW. Since the UPE is bridging-capable, only a
single PW is required per VPLS instance for any number of access connections in the same
VPLS service. This reduces the signaling overhead between UPEs and NPEs.

NPE Operation
An NPE device supports all bridging functions for VPLS service and supports the routing and
MPLS encapsulation as in basic VPLS. The operation of the NPE is independent of the type of
device at the other end of the spoke PW. Thus the NPE will switch traffic between the spoke
PW, hub PWs, and attachment circuits once it has learned the MAC addresses.

4-64 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Dual-Homed UPE
The failure of an NPE router or the connection of the PW can cause the UPE to suffer a total
loss of connectivity. To prevent this, redundant connections can be provided. The UPE is dual-
homed into two NPE routers. One ofthe two PWs is designated as primary and the other as
standby. The UPEs negotiate PW labels for both PWs, but do not use the standby PW unless
the primary PW fails. Spanning-tree instance or manual configuration can be used to designate
primary and standby status.
Upon failure of the primary PW, the UPE immediately switches to the standby PW. At this
point, the NPE that terminates the standby PW starts learning MAC addresses on the spoke
PW. All other NPEs initially continue to send traffic to the initial NPE until they learn that the
devices are now connected to a new NPE. To enable a faster unlearning process, the new NPE
may send out a flush message using the MAC list type, length, and value (TLV) (type Ox404) to
all NPEs. Upon receiving the message, the NPEs flush the MAC addresses associated with that
VPLS instance.

- Flat VPLS HVPLS

Pros Simple provisioning Suitable for large environments

Reduced replication and signaling
overhead on NPEs
Expansion affects new nodes only

Cons Scalability limitation to small More complicated provisioning

environments More complex design and
PE packet replication operations
Directed LDP full mesh - More expensive hardware for
n * (n-1 )/2 sessions MPLS-based access
iCl2012Ciscoandioritsafflllllte!l.A1I~ghts",served.

Flat VPLS offers the easiest implementation method. It supports both port and VLAN transport
modes, but is limited to small deployments. All packet replication is performed on the PE
devices. It results in a directed LDP full mesh ofn*(n-l)/2 sessions.
H-VPLS is suitable for larger environments. It places less replication burden and signaling
overhead on the NPE routers. It allows the service provider to expand certain network areas
without affecting the core and most of the NPE infrastructure. It requires, however, more
complex design, provisioning, and operations, and may incur higher cost, if MPLS is deployed
in the edge.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-65
 Requires special address family for MP-BGP neighbors
- address-family 12vpn vpls-vpws
Available for VPLS and VPWS
Two signaling methods: LOP and BGP
- Both methods use VLAN IDs, ROs, and RTs to limit discovery scope.

CE

PE 111:..... PE

PW full mesh can be

autodiscovered.

CE PE CE

C2012Ciscoandioritsafflllllte!l.AII~ghts",served.

Once the PEs have ascertained that other PEs have an association with the same VPLS
instance, each PE needs to set up a PW between themselves and bind the PWs to the particular
VSI. These PWs can be set up using one of these methods:
For neighbor autodiscovery:
Static configuration
BGP-based autodiscovery

For PW signaling:
BGP-based signaling
LDP-based signaling

LOP-Based Signaling
LDP is used for the signaling of PWs that are used to interconnect the VPLS instance of a given
customer on the PEs. In order to signal a full mesh ofPWs, a full mesh of Targeted LDP (T-
LDP) sessions is required between the PEs. In the absence of autodiscovery, these sessions
must be manually configured on each PE router. This LDP session is used to communicate the
inner label or ve label that must be used for each PW. The network operator assigns aVe ID
that is used to identify a particular PW, and it is configured to be the same for a particular
VPLS instance on all PE routers.
The LDP session communicates the label value and the FEe element that includes the ve ID,
control word bit (indication whether a control word will be used), ve type (Ethernet or VLAN-
tagged Ethernet), and interface parameters field (media maximum transmission unit (MTU),
and so on.)

4-66 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 LDP signals the setup, maintenance, and teardown of a PW between PE devices. Once a PE
discovers that other PEs have an association for a particular VPLS instance, the PEs signal
using T-LDP to other PEs that a PW is required to be set up between PEs. When a PE device is
associated with a particular VSI, LDP transmits a label-mapping message with VC type Ox0005
and a 4-byte VC ID value. If the remote PE has an association with that particular VC ID, it
will accept the LDP label-mapping message and respond with its own label-mapping message.
Once the two unidirectional VCs are operational, they are combined to form a bidirectional
PW.

BGP-Based Autodiscovery and Signaling

The BGP Network Layer Reachability Information (NLRI) takes care of autodiscovery and
signaling at the same time. The NLRI is generated by a given PE containing the necessary
information required by any other PE. These components enable the automatic setting up of a
full mesh ofPWs for each VPLS.
On each PE, a route distinguisher (RD) and a route target (RT) are configured for each VPLS,
like in Layer 3 VPN and Layer 2 VPN. The RT is same for a particular VPLS across all PEs,
and is used to identify which VPLS a particular BGP message pertains to. The RD is used to
disambiguate routes. On each PE, for each VPLS, an identifier is configured called a VPLS
Edge Identifier (VE ID). Each PE involved in a particular VPLS must be configured with a
different VE ID. BGP is used to advertise the VE ID to other PEs. This, along with other
information in NLRI, is used to calculate the value of the PW label required to reach the
advertising PE.
For a given VPLS, a PE requires that each remote PE uses a different PW label to send traffic
to that PE. This requirement is done to facilitate the MAC learning process. This way, the
receiving PE can learn which PE is associated with the source MAC address of the frame. APE
could send a list ofPW labels that are required to reach it, one per remote VPLS edge in that
VPLS. However, this would mean that a PE sends a long list of labels if there is a large number
of PEs. Instead, the necessary information is carried in Border Gateway Protocol (BGP) NLRI.
This allows all remote PEs to calculate the PW label expected by the advertising PE.
A BGP update message contains the extended community RT (which allows the receiving PE
to ascertain which particular VPN the advertisement pertains to), Layer 2 information (control
flags to indicate whether a control word is included, the encapsulation type-Ethernet or
VLAN-tagged-and MTU), and other BGP attributes such as autonomous system (AS) path,
origin, and so on.
The NLRI contains the RD, VE ID, label base, VE block offset, and VE block size. The label
base, VE block offset and VE block size are the information that the receiving PE requires to
calculate the PW label when sending traffic for VPLS to the advertising PE.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-67
Implementing VPLS and H-VPLS
This topic describes the implementation ofVPLS and H-VPLS.

Prepare MPLS infrastructure:

- P E routers must have a /32 address on their loop backs.
- PE loopback addresses cannot be summarized in the core.
- Ensure MTU sizes in the core are large enough.
Enable Layer2 frame transport on both endpoint attachment circuits.
Make sure MTU is the sameon both endpointinterfaces.
Configure bridge group and bridge domain.
Assign interface(s) to the bridge domain.
Configure VFI with statically defined PWs or neighbor autodiscovery.

To implement VPLS or H-VPLS, you first need to prepare the MPLS infrastructure. This
process includes these steps:
PE routers must have a /32 address on their loopbacks. This is required for binding the VC
label and the transport LDP into a label stack.
PE loopback addresses cannot be summarized in the core. Aggregation in the network
would break the LSP path.
MPLS must be enabled in the core.
MTU sizes on the core links must be able to accommodate the customer MTU with the
added label overhead.

Once the MPLS infrastructure is ready, you will do the following:

Enable Layer 2 frame transport on both endpoint PE routers.
Make sure the MTU is the same on both endpoint interfaces.
Configure the bridge group and bridge domain.
Assign interfaces to bridge domain.
Configure the VFI with either statically defined PWs or neighbor autodiscovery.

4-68 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 PE1: PE2:

interface LoopbackO interface LoopbackO

ipv4 address 10.1.1.1255.255.255.255 ipv4 address 10.2.1.1255.255.255.255

interface Gigabi tEthernet 0 /0/0/0.10 interface GigabitEthernetOjOjOjO .10

12transport 12 transport
encapsulation dotlq 10 encapsulat ion do tlq 10
!
12vpn 12vpn
bridge group VPLS -groupl bridge group VPLS-groupl
bridge-domain VPLS- domainl bridge-domain VPLS-domainl
interface GigabitEthernetOjOjOjO .10 interface GigabitEthernetOjOjOjO.10
exit exit
vfi VPLS-vfil vfi VPLS-vfil
neighbor 10.2.1.1 pw-id 4 neighbor 10.1.1.1 pw-id 4
neighbor 10.3.1.1 pw-id 6 neighbor 10.3.1.1 pw-id 8

The first scenario is a flat VPLS implemented on Cisco lOS XR platforms.

The attachment circuits on both ends are enabled for Layer 2 transport using the 12transport
keyword.
The VPLS is configured within the Layer 2 VPN (12vpn command), bridge group, and bridge
domain configuration mode. The bridge domain contains two main configuration elements: a
list of the local attachment circuits, and the VFI that contains all required PWs.
The names of the bridge group and domain have local significance. The PW ID must match on
both ends.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-69
 PE1: PE2:

interface LoopbackO interface LoopbackO

ipv4 address 10.1.1.1 255.255.255.255 ipv4 address 10.2.1.1255.255.255.255

interface Gigabi tEthernet 0 /0/0/0.10 interface GigabitEthernetOjOjOjO .30

12transport 12 transport
encapsulation dotlq 10 encapsulation dotlq 30
rewrite ingress tag translate l-to-l rewrite ingress tag translate l-to-1
dotlq 99 symmetric dotlq 99 symmetric

12vpn 12vpn
bridge group VPLS -groupl bridge group VPLS-group3
bridge-domain VPLS- domainl bridge-domain VPLS-domain3
interface GigabitEthernetOjOjOjO .10 interface GigabitEthernetOjOjOjO.30
exit exit
vfi VPLS-vfil vfi VPLS-vfi3
neighbor 10.2.1.1 pw-id 4 neighbor 10.1.1.1 pw-id 4
neighbor 10.3.1.1 pw-id 6 neighbor 10.3.1.1 pw-id 8

This scenario differs from the previous one in managing the outmost VLAN tag. In this
example, the VLAN tag is replaced with the VLAN tag 99. On the egress PE, the VLAN tag 99
will be swapped with the local VLAN tag 30.
The service provider will use rewrite because the service provider offers some service on
VLAN 99, and customers uses different VLANs.
This technique illustrates how the service provider can replace customer VLAN tags with a tag
from its own range for transport across the core network.

4-70 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 .-... -5
FaO/1
P-SW1

1,1 I~ol
interface Fas tEthernetO /1
description CE-SW interface LoopbackO
switchport access vlan 10 ipv4 address 10.1.1.1 255.255.255.255
swi tchpor t mode dotl q- tunne 1
interface GigabitEthernetOjOjOjO.lO 12transport
interface Fas tEthernetO /2 encapsulation dotlq 10 second-dotlq any
description N-PE rewrite ingress tag translate I-to-l dotlq 99
swi tchport mode trunk symmetric

12vpn
br idge group VPLS -groupl
bridge-domain VPLS-domainl
interface GigabitEthernetOjOjOjO.lO
exit
vfi VPLS-vfil
neighbor 10.2.1.1 pw-id 4
neighbor 10.3.1.1 pw-id 6

H-VPLS requires that either QinQ or MPLS is used in the access network. This scenario
depicts QinQ access.
The switch must be configured for QinQ tunneling. This requires the switchport mode dotlq-
tunnel and the switchport access vlan commands on the CE-facing port. The switchport mode
dotlq-tunnel command instructs the switch to encapsulate incoming 802.1 Q frames within the
VLAN tag that is specified with the switchport access vlan command. In this case, P-SW1
applies the outer VLAN tag 10 to all frames that are received from the customer switch CE-
SWI.
The PE routers are configured for regular VPLS with static or autodiscovered PWs. The
encapsulation on PEl is set using the command encapsulation dotlq 10 second-dotlq any to
match only QinQ frames and exclude 802.1 Q frames. The command encapsulation dotlq 10
would also work in this scenario. In this case, PEl translates the outer VLAN tag to 99 for
transport to the egress PE.
PE2 receives the VPLS packets and translates the outer VLAN tag 99 to the VLAN tag
allocated to the local customer, 30. This is achieved with the rewrite ingress tag translate I-
to-I dotlq 99 symmetric command along with the encapsulation dotlq 30 second-dotlq any
command.
The configurations ofPE2 and P-SW2 are configured in the same way, symmetrically to PEl
and P-SWl, except that the customer frames are encapsulated in VLAN tag 30 instead of 10.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-71
 RP/O/RSPO /CPUO :PE3# show l2vpn bridge-domain detail
Sat Nov 26 13:48:47.127 UTC
Bridge group: VPLS-group3, bridge-domain: VPLS-domain3, id: 1, state: up, ShgId: 0,
MSTi: 0
MAC learning: enabled
MAC withdraw: enabled
MAC withdraw for Access PW: enabled
Flooding:
Broadcast & Multicast: enabled
Unknown unicast: enabled
MAC aging time: 300 s, Type: inactivity
MAC limit: 4000, Action: none, Notification: syslog
MAC limit reached: no
MAC port down flush: enabled
MAC Secure: disabled, Logging: disabled
Split Horizon Group: none
Dynamic ARP Inspection: disabled, Logging: disabled
IP Source Guard: disabled, Logging: disabled
DHCPv4 snooping: disabled
IGMP Snooping prof ile: none
Bridge MTU: 1500
MIB cvplsConfigIndex: 2
Filter MAC addresses:
Create time: 26/11/201111:38:38 (02:10:08 ago)
No status change since creation
ACs: 1 (1 up), VFIs: 1, PWs: 1 (1 up), PBBs: 0 (0 up)
< to be continued>

The show 12vpn bridge-domain detail command is used to verify the VPLS operation. It
provides a verbose output that is only partly shown here. This section describes parameters that
relate to bridging, MAC learning, and security features active with the particular VPLS domain

Lis t of ACs:
AC: GigabitEthernetO/O/O/O .30, state is up
Type VLAN; Num Ranges: 1
VLANranges: [30,30]
MTU 1504; XC ID Ox840001; interworking none
MAC learning: enabled
Flooding:
Broadcast & Multicast: enabled
Unknown unicast: enabled
MAC aging time: 300 s, Type: inactivity
MAC limit: 4000, Action: none, Notification: syslog
MAC limit reached: no
MAC port down flush: enabled
MAC Secure: disabled, Logging: disabled
Split Horizon Group: none
Dynamic ARP Inspection: disabled, Logging: disabled
IP Source Guard: disabled, Logging: disabled
DHCPv4 snooping: disabled
IGMP Snooping profile: none
Storm Control: disabled
Static MAC addresses:
Statistics:
packets: received 31686, sent 27420
bytes: received 2156476, sent 1911176
Storm control drop counters:
packets: broadcast 0, multicast 0, unknown unicast
bytes: broadcast 0, multicast 0, unknown unicast 0
<to be continued>

The next part of the output describes all attachment circuits assigned to the bridge domain. It
provides information on enabled security features and various statistics.

4-72 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
 List of Access PWs:
List of VFIs:
VFI VPLS-vfi3
PW: neighbor 10.7.1.1, PW ID 64500: 10, state is up ( established
PW class not set, xc ID Oxfffc0005
Encapsulation MPLS, Auto-discovered (BGP), protocol LDP
PW t'YPe Ethernet, control word disabled, interworking none
PW backup disable delay 0 sec
Sequencing not set

MPLS Local Remote

Label 30000 16002

BGP Peer ID 10.3.1.1 10.7.1.1
LDP ID 10.3.1.1 10.7.1.1
AIl 10.3.1.1 10.7.1.1
AGI 64500,10 64500,10
Group ID Ox1 Ox1
Interface VPLS -vfi3 VPLS-vfi7
MTU 1500 1500
Control word disabled disabled
PW t'YPe Ethernet Ethernet
VCCV CV type Ox2 Ox2
(LSP ping verification) (LSP ping verification)
vccv CC type Ox6 Ox6
(router alert label) (router alert label)
(TTL expiry) (TTL expiry)

<ou tpu t t runca ted>

The last major part of the output focuses on the PWs that constitute the VFI. In this section, you
can view the PW parameters on the local side and on the remote end.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-73
Summary
This topic summarizes the primary points that were discussed in this lesson.

VPLS neighbors can be configured manuallyor learned by the

autodiscovery process .
VPLS is implemented using bridge groups, bridge domains, and VFls.

4-74 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Summary
This topic summarizes the primary points that were discussed in this module.

3
Layer 2 VPNs are classified as VPWS (point-to-point) and VPLS
(point-to-m ultipoint).
EoMPLS is the most common AToM method that supports a host of
features, such as inter-AS operation, redundancy, and EVC
infrastructure.
VPLS configuration uses VFls for bridging over the pseudowires and
can be combined with Cisco EVC capabilities for matching and
manipulating frames.

~2012Cioco.ndlor~.afl'ili_.Allrighlol. . .rv"'.

Layer 2 VPNs are classified as Virtual Private Wire Service (VPWS) (point-to-point) and
Virtual Private LAN Services (VPLS) (point-to-multipoint). They can be transported over
Multiprotocol Label Switching (MPLS) or IP networks. VPLS can only be carried over MPLS
and cannot use Layer 2 Tunneling Protocol version 3 (L2TPv3), which provides the transport
protocol in the pure IP core.
Any Transport over MPLS (AToM) is a subset ofVPWS. It is deployed to carry Ethernet,
Frame Relay, PPP and High-Level Data Link Control (HDLC) frames, and ATM cells, and it
emulates time-division multiplexing (TDM) circuits over MPLS. It supports interworking of
disparate encapsulations at both ends and allows the connections to span multiple autonomous
systems.
AToM requires that MPLS and Targeted Label Distribution Protocol (T-LDP) sessions are
operational, which includes that the loopback addresses on PE routers use a /32 network mask
and are not summarized in the core. Maximum transmission unit (MTU) considerations include
the requirement of having identical MTU sizes on both sides, and guaranteeing sufficiently
large MTU sizes in the core to avoid fragmentation. Fragmentation is not supported for Layer 2
VPNs.
Carrier Ethernet offers a range of enhancements for scalability (802.lad and 802.1ah), high
availability (MST, REP, and G.8032), and Operation, Administration, and Maintenance (OAM)
(802. lag and link OAM).
VPLS configuration uses virtual forwarding instances (VFIs) for bridging over the pseudowires
(PWs) and can be combined with Cisco Ethernet Virtual Connection (EVC) capabilities for
matching and manipulating frames.

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-75
References
For additional information, refer to these resources:
To learn more about implementing Layer 2 VPN on Cisco lOS XR platforms, refer to
Implementing Virtual Private LAN Services on Cisco IDS XR Software at this URL:
http://www.cisco.com/enIUS/docs/routers/crs/software/crsJ3. 9/lxvpn/configuration/guide/
vc39vpls.pdf
To learn more about configuring Layer 2 VPNs and Ethernet services on Cisco lOS XR
platforms, refer to Cisco ASR 9000 Series Aggregation Services Router L2VPN and
Ethernet Services Configuration Guide at this URL:
http://www.cisco.com/enIUS/docs/routers/asr9000/software/asr9kJ4.1/lxvpn/configuration
/guide/lesc41.html

4-76 Implementing Cisco Service Provider Next-Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.
Module Self-Check
Use the questions here to review what you learned in this module. The correct answers and
solutions are found in the Module Self-Check Answer Key.
QI) To establish VPWS in your service provider environment, MPLS must be configured in
your IP core network. (Source: Introducing Layer 2 VPNs)
A) true
B) false
Q2) To establish VPLS in your service provider environment, MPLS must be configured in
your IP core network. (Source: Introducing Layer 2 VPNs)
A) true
B) false
Q3) A IOOO-byte IPv6 packet is transported over EoMPLS. The control word is enabled.
MPLS network is not configured for traffic engineering. The Layer 2 PDU of that
packet in the core is _ bytes. (Source: Introducing AToM)
Q4) A QinQ frame with outer VLAN tag 10 and inner VLAN tag 99 is transported over
VPLS. The ingress PE has two interfaces with these settings:
encapsulation dotlq 10, rewrite ingress tag translate 1-to-1 dotlq 70
symmetric
encapsulation dotlq 10 second 10-200, rewrite ingress tag pop 1 symmetric
The frame transported over VPLS has this VLAN tagging: (Source:
Implementing VPLS)

2012 Cisco Systems, Inc. Layer 2 VPNs and Ethernet Services 4-77
Module Self-Check Answer Key
Ql) B
Q2) A
Q3) 1012
Q4) 99 (only one VLAN tag)

4-78 Implementing Cisco Service Provider Next Generation Edge Network Services (SPEDGE) v1.0 2012 Cisco Systems, Inc.