White Paper The Benefits Of Migrating to MPLS

www.xo.com c onne c t / c ommunic ate / ma nage / p rote c t / optimize

The Benefits of Migrating to an MPLS-based Private IP Network:

Contents

Introduction 3
The Difference in MPLS-based Private and Site-to-Site IP VPNS 3
Site-to-Site IP VPNs 4
MPLS-based Private IP VPNs 4
The Benefits of Private IP Networks 5
Class of Service 6
Automatic Redundancy/Disaster Recovery 7
Fully Meshed Infrastructure 7
Reduced Complexity 8
Getting Started 8
Prioritizing Applications 8
Managing Class-of-Service Thresholds 9
Measuring Service Level Agreements 9
Optimizing Bandwidth Requirements 9
Managing the Network Migration 9
Overcoming Private IP Challenges 10
Prioritizing Applications 10
Managing Class-of-Service Thresholds 13
Measuring Service Level Agreements 14
Optimizing Bandwidth Requirements 14
Managing the Network Migration 14
Conclusion 15
About XO MPLS & Application Performance Management 16
About XO Communications 16

2 Solutions Built Around You.

 XO Communications

Introduction

Today’s enterprises are faced with a number of network-related needs that are challenging
from both a technology and a financial perspective. Organizations are looking for proven,
cost-effective solutions to:

• Create and improve disaster recovery infrastructures

• Replace outdated hub-and-spoke architectures
“Private IP networks • Meet increasingly complex network requirements
(generally MPLSbased, but • Prioritize applications, such as VoIP
• Maximize performance while minimizing costs
not exclusively) are rapidly
growing in popularity for Your enterprise might have one or many of the needs listed above and may be in the midst of
evaluating solutions to meet your goals and objectives. One of today’s hot topics for network
medium- to large-sized
and application executives to help meet these growing demands is Multi-Protocol Label
enterprises across a wide range Switching (MPLS) based/private IP networks. Private IP networks—generally MPLS-based,
of industries.” but not exclusively—are rapidly growing in popularity for medium- to large-sized enterprises
across a wide range of industries. Benefits for adopting these networks include the ability to
prioritize bandwidth by application, the potential for automatic redundancy, and the value of a
fully meshed network with less complexity and cost.

However, before you begin migrating your existing network to a private IP network, consider
these challenges of MPLS-based networks:

• What applications you are running

• How to prioritize them
• The need for visibility into individual IP subnet pairs.

Many white papers describe technically how MPLS works—down to the details of
configuration across different Customer Premises Equipment (CPE) types but, before you
take a deep dive into the inner workings of the technology, determine if migrating to MPLS is
the best decision for your enterprise. Thoroughly investigate the key business reasons your
enterprise might consider a private IP network, as well as the issues you must monitor and
manage to make MPLS a success.

The Difference in MPLS-based Private and Site-to-Site IP VPNS

Before discussing private IP networks in more detail, let us first appreciate the number of
different flavors of IP Virtual Private Networks (VPNs). The term IP VPN has many different
connotations, but in general, IP VPNs are broken into two distinct categories: site-to-site and
MPLS-based/private IP VPNs.

Connect / Communicate / Manage / Protect / Optimize 3

The Benefits of Migrating to MPLS

Site-to-Site IP VPNs

A site-to-site IP VPN, also known as a CPE-based VPN, is defined as using the public
Internet as the core backbone of the VPN. Enterprises use dedicated Internet connectivity,
cable, wireless, DSL, satellite, and dial-up as different means to connect to the IP VPN.
Enterprises then typically use CPE—VPN gateway, router, etc.—to create a VPN with
tunnels and encryption so different remote sites can communicate with the headquarters
and with other remote sites directly and securely.

The primary benefit for site-to-site IP VPNs is tied to access—low cost and ubiquity.
Traditionally, Internet connectivity is much less expensive than frame relay or ATM—and
“While there are cost savings the difference grows quite substantially for international locations. The ubiquitous nature
to access site-to-site IP VPNs, of Internet connectivity is a major benefit as well—IP is IP regardless of whether it is
the total cost of ownership dial-up, dedicated, domestic, or international.

can actually be higher for While there are cost savings to access site-to-site IP VPNs, the total cost of ownership
enterprises because CPE is can actually be higher for enterprises because CPE is needed at every site to create
needed at every site to create the VPN. The potential exposure of using the public Internet for business-critical traffic
is extremely high. The Internet’s architecture was not designed for business-grade
the VPN.” applications and requirements and does not offer strong Service Level Agreements (SLAs)
for enterprises.

MPLS-based Private IP VPNs

An MPLS-based private IP VPN, also known as a network-based VPN, is defined as using

a service provider’s offering—usually frame relay and ATM and with more frequency,
Ethernet—as the backbone of the VPN. MPLS is a solution that allows the service
provider’s core routers to transition from Layer 2-based connectivity to Layer 3-based
connectivity.

For example, traditional frame relay requires a dedicated permanent virtual circuit (PVC)
for remote sites to communicate between each other, as illustrated in Figure 1. The traffic
follows the path for every transaction between two sites.

Figure1: Frame Relay or ATM Network

4 Solutions Built Around You.

 XO Communications

With a private, IP-enabled transport network—as illustrated in Figure 2—the carrier’s routers
use MPLS-based routing to communicate between locations in the network via IP addressing
and is not limited to individual PVCs. Tunneling and encryption are generally not required
for the enterprise because the network is on the private carrier’s backbone, not the public
Internet. Additionally, the carrier’s core routers create the VPN, not the enterprise’s CPE.

“A key advantage of a
private IP VPN is the ease of
migration from a frame relay
or ATM network.”

Figure 2: IP-enabled Transport Network

A key advantage of a private IP VPN is the ease of migration from a frame relay or ATM
network. For most service providers, there is no need for provisioning new circuits—since the
frame relay and ATM is still the Layer 2 access—and routers may not need to be replaced.
The complexity to the enterprise of managing the VPN is less with private IP networks
because it is network based, meaning the provider, by means of its core routers, handles the
creation of the VPN.

The frame relay world considers two bandwidth numbers: port size and committed
information rate (CIR). The private IP world manages two criteria: Committed Access Rate
(CAR) and thresholds by Class of Service (CoS). The frame relay port size and private IP
CAR are very similar—this is the maximum amount of bandwidth that can be used for a single
circuit/location. The differences lie between CIR and CoS.

Enterprises purchase CIR per PVC as a way to guarantee bandwidth from the service
provider. Private IP VPNs provide a CoS capability to the enterprise by allowing multiple
classes of prioritization, each a percentage of CAR. CoS is more application-focused since
enterprises can select which applications are the most critical or the most delay-sensitive
and assign them the highest priority. The network-based VPN polices and queues the classes
instead of deploying traffic shapers to prioritize the traffic.

The Benefits of Private IP Networks

Assuming your enterprise already has a Wide Area Network (WAN), the question most people
ask themselves is why migrate an existing network that is working fine to a new, unknown
platform or technology?

Connect / Communicate / Manage / Protect / Optimize 5

The Benefits of Migrating to MPLS

Any network manager working 10 years ago or so knows why enterprises migrated
from point-to-point/private line networks to frame relay—it was cheaper. While there
will likely be cost savings for access by migrating to private IP, there are several other
key benefits of moving from a frame relay or ATM network to a private IP VPN. These
benefits include:

• Class of Service (CoS)

• Automatic redundancy/disaster recovery
“A “While there will likely • Fully meshed infrastructure
• Reduced complexity
be cost savings for access by
migrating to private IP, there Class of Service
are several other key benets of Prioritizing applications by CoS has been the key driver for many enterprises rolling
moving from a frame relay or out private IP networks. Instead of using critical router resources to shape traffic or
ATM network to a private IP buying an expensive traffic shaper to prioritize your applications, the service provider’s
core routers do the heavy lifting in prioritizing different classes of service in private IP
VPN.” networks.

Enterprises decide which applications are the most mission-critical or delay sensitive
and assign them specific classes. Most service providers today offer between three and
six different classes of service. They might have different marketing names, but the key
point is they differentiate policies for the applications to travel through the network—
each class has a different priority setting than the others. In addition to the number of
classes, the provider also allows a certain amount of bandwidth per class—depending
on total circuit size, number of classes and contract agreement. Once you know the
number of classes and the amount of bandwidth assigned to each, you are ready to
begin assigning priority.

For your enterprise, Voice over IP (VoIP) and Oracle® might be the most important
applications, so they are set for the highest priority, while e-mail and Web browsing
receive lower assignments. You configure your router to tag the correct class in the IP
header—DiffServ or TOS bit settings—and pass the data to the service provider’s edge
router. The provider’s router looks at the IP header for the class setting and polices
the traffic. The policing from the source site—or ingress—ensures the threshold of
bandwidth is not exceeded for each class setting. The core router then sends the traffic
across the network with prioritization.

At the destination location—or egress— the provider’s router queues the prioritized
traffic from each site and delivers the highest priority first. With the capability of CoS,
you can easily implement a prioritization schedule across the enterprise so the most
mission-critical applications receive the highest priority.

6 Solutions Built Around You.

 XO Communications

Automatic Redundancy/Disaster Recovery

The need for infrastructure redundancy has grown substantially over the past few years as
more enterprises drive revenue, reduce costs, or provide services based on applications on
the WAN. If a credit card authorization site is down, a retail enterprise may lose sales because
today’s consumers carry less cash. If a manufacturing company cannot transmit its stocking
order to a plant, the vendor might impose penalties. These two examples are common
scenarios that are exacerbated when there is a single point of failure in the infrastructure.
“A “The need for That does not always mean the network itself is the cause of the problem; it can be any
infrastructure redundancy portion of the infrastructure that does not allow the business transaction.

has grown substantially over

If a transaction cannot be completed at the primary location, enterprises can forward it
the past few years as more for completion to a back-up location. In a frame relay or ATM environment, this is typically
enterprises drive revenue, handled by provisioning additional PVCs between critical sites. While this architecture is
viable, it can be extremely expensive, depending on the size and scope of the network.
reduce costs, or provide
Managing tens or hundreds of PVCs at multiple sites can be a logistical and managerial
services based on applications nightmare–especially at remote locations with no networking support staff. Enterprises must
on the WAN.” determine if the exposure to lost revenues and incurred additional costs exceeds the extra
resources and complexity for a disaster recovery infrastructure.

A private IP network is fully redundant, based on its Layer 3-based connectivity—IP

subnet-to-IP subnet. Now each site in your entire network has access to every other site.
Implementing a disaster recovery/back-up strategy is much easier and more cost-effective
since dedicated PVCs are no longer needed for each and every location.

Fully Meshed Infrastructure

In a hub-and-spoke environment, traffic from a remote site must traverse a host location
before it is routed to the destination address. This architecture was adequate several years
ago, but as Web-based applications or time-sensitive applications like VoIP become more
prevalent, the hub-and-spoke architecture becomes more stressed.

A fully meshed architecture—meaning every site can communicate directly with any other
site without having to run through a hub/host location first—has two key benefits for most
enterprises: improved site-to-site performance and fewer burdens on host locations.

When an application such as VoIP is used between two remote sales offices, there is no
benefit to “home-running” the application back to the host location. This scenario adds
potential delay by adding extra steps and distance to complete the application transaction.
With a fully meshed network, a VoIP call between two offices in London will flow directly
between the locations, instead of having to go to the hub site located in Los Angeles. In a
private IP network, the number of steps and physical distance alone can be greatly reduced.

Connect / Communicate / Manage / Protect / Optimize 7

The Benefits of Migrating to MPLS

In conjunction with improving parameters with a fully meshed network—including delay or

jitter—reducing host site bandwidth usage is also a major benefit. Since the organization does
not bring all traffic back to the host site, this may greatly reduce the bandwidth requirements
at a host location, as well as enhance the performance across the entire infrastructure.

Reduced Complexity

The complexity of managing a frame relay or ATM network grows exponentially with network
size and the number of virtual circuits. Managing hundreds of sites and thousands of PVCs
is a daunting task for many enterprises just to handle moves, adds, and changes on a daily
basis.

“Managing hundreds of sites As discussed earlier, IP subnet addressing is used to connect every site in the network. PVCs
and thousands of PVCs is are no longer the connection between sites. Instead of managing a port for every site and
a daunting task for many tens, hundreds, or thousands of individual PVCs, the private IP network has a single port
for each site and then uses IP addressing to connect to every other site. This architecture is
enterprises just to handle much less complex, meaning it is easier to administer and allows enterprises to focus limited
moves, adds, and changes on resources on more important activities.
a daily basis.”
Getting Started
Every enterprise can likely benefit in some fashion from a private IP network. Before deciding
to migrate your existing frame relay or ATM network to private IP, consider these key factors
to achieve maximum benefit with minimum pain:

• Prioritizing applications
• Managing CoS thresholds
• Measuring SLAs across multiple classes
• Optimizing bandwidth requirements
• Managing the network migration

Prioritizing Applications

Most enterprises know their most business-critical applications either by management

direction, trial and error, or which ones get the most complaints if there are performance
issues. But after the first handful of applications, many enterprises have difficulty slotting
other applications in medium or low classes. Compounding the problem is a lack of knowing
what applications are actually on the network. If you do not know what applications are on
your network, it is impossible to prioritize them.

Another challenge for network managers is the addition of new, powerful applications. If the
enterprise decides to implement Oracle or SAP in the next six months, how will the new high-
priority application affect existing high-priority applications? Networks and applications are
continuously evolving and changing, and CoS prioritization must be able to adapt.

XO911WP-1010

8 Solutions Built Around You.

 XO Communications

Managing Class-of-Service Thresholds

Because you are now prioritizing applications with more granularity, managing CoS
thresholds becomes absolutely critical. As mentioned earlier, service providers allow a
certain amount of bandwidth for class of service. As long as you stay below that usage
threshold, there should not be any problems. If you exceed the threshold, the pain can be
magnified. If the network is congested and exceeds your highest priority class threshold,
the traffic will drop to the lowest priority or may be discarded. So now your most mission-
critical applications may be at risk. Rolling out new applications across the infrastructure
compounds the problem.

A new application will be assigned a class, and its usage may cause a domino effect for
other applications. For example, a new custom financial application might cause the highest
priority class to exceed the threshold, so one of the other applications in the same class
will be dropped one prioritization lower, and so on. As an enterprise with mission-critical
applications, you cannot risk the highest priority applications being downgraded because the
class threshold was exceeded.
“In today’s world of squeezed
budgets, optimizing Measuring Service Level Agreements
bandwidth is critical for Many enterprises have SLAs with the service provider to ensure the network can handle its
many enterprises..” applications. Measuring SLAs from a provider’s point of view is delivering a monthly report
with weighted averages across the entire infrastructure for delay, throughput, and availability.
Private IP compounds the issue by eliminating traditional virtual circuits, as well as layering
on multiple classes of service.

Before private IP, it was difficult for many enterprises to proactively monitor service level
parameters from an end-to-end point of view. Now with the extra complexity, some
enterprises find it extremely difficult to measure SLA parameters across IP subnets and by
individual classes of service. This lack of visibility makes it more difficult to leverage the day-
to-day benefits of SLAs.

Optimizing Bandwidth Requirements

With today’s limited budgets, optimizing bandwidth is critical for many enterprises. While
enterprises need to ensure they have sufficient bandwidth to meet the needs of applications
and end-users, they do not have the luxury of over-provisioning every circuit in the network.
But as application performance becomes more important, the enterprise cannot be so
aggressive and not have sufficient bandwidth resources. Most organizations walk this fine
line.

Managing the Network Migration

The ease in migrating from frame relay or ATM to private IP was highlighted earlier in this
paper. This migration becomes easy because organizations can leverage existing Layer
2 infrastructure and CPE. Enterprises cannot let that ease lull them into a false sense of
security as they plan migration efforts.

Connect / Communicate / Manage / Protect / Optimize 9

The Benefits of Migrating to MPLS

Even though it may be a fairly simple migration path to private IP, the transition takes months.
You will not be able to have 75 sites converted from frame relay to private IP overnight. You
must manage your network throughout the entire migration. Some of your locations may
be frame relay while others are already converted to private IP. Without visibility into the
performance of both, you will be managing blind.

Overcoming Private IP Challenges

Enterprises are seeing the benefits of a private IP infrastructure but remain worried about
some or all of the challenges we highlighted. You are likely weighing the benefits of private IP,
but you may be uncertain whether you should migrate from your existing infrastructure—one
that is working for you.
“Even though it may be a
Some people follow the old adage, “if it isn’t broke, don’t fix it.” Sticking strictly to that
fairly simple migration path mentality, you would still be using carbon paper to make copies or pen and paper to balance
to private IP, the transition the ledger of a billion-dollar organization. In a networking context, it would be like building a
takes months. You will not 300-site network with purely dedicated circuits. That strategy would still work today, but it
probably would not be the most cost efficient or robust. You need to find a way to eliminate
be able to snap your fingers
the challenges of migrating to private IP so you can reap its benefits.
and have 75 sites converted
from frame relay to private IP
Prioritizing Applications
overnight.”
Providers, such as XO Communications, offer Applications Performance Management
(APM) tools to help identify and solve the two critical aspects of prioritizing applications:
what applications are on the network and whether the applications are prioritized correctly.
The XO Performance Management tool auto-discovers the applications across the entire
infrastructure (see Figure 3). So instead of prioritizing a subset of all the applications, you now
have the knowledge of total application usage so you can completely assign the CoS settings.
Once users prioritize applications, they determine if each has been properly configured.
Two potential pitfalls commonly occur during this process: IP header configurations are set
incorrectly and applications are placed in the wrong class.

10 Solutions Built Around You.

 XO Communications

“With XO Applications
Performance Management,
identifying misconfigured
applications is extremely
fast and easy by measuring
utilization for each CoS across
every site in the network.”

Figure 3: Applications Infrastructure

The main issue for many enterprises that have already migrated to private IP is application
misconfiguration. For one enterprise example, VoIP was deemed critical and should have
received the highest-class setting. However, as the network engineer was configuring the
80 sites, he made a simple mistake when setting the DiffServ for one location and VoIP
calls were impacted. Without visibility, identifying the problem was extremely difficult.
They thought VoIP was receiving the highest priority, so they began troubleshooting other
parameters including CPE.

Connect / Communicate / Manage / Protect / Optimize 11

The Benefits of Migrating to MPLS

Figure 4: Measure usage for each CoS setting, including

“Network managers must unknown/misconfigured applications

take steps to ensure that

With an APM tool, the network engineer was able to look at the site impacted by
applications are assigned poor VoIP quality and could quickly identify that a significant amount of traffic was
correctly.” categorized as unknown (see Figure 4). Using deep-packet inspection and the traffic
capture feature, the misconfigured application can be identified in a matter of minutes.

Figure 5: See which applications are using every class of service

12 Solutions Built Around You.

 XO Communications

Now that the applications across the entire network have been identified and each IP header
configuration has been set correctly, you need to verify that each application is set correctly.
You can see that the highest setting has applications including Oracle, routing, and Web
traffic (see Figure 5).

Network managers must take steps to ensure that applications are assigned correctly. In this
example, Web traffic should have been assigned a lower priority, but it is consuming valuable
high-priority resources. Armed with this information, the end-user can correctly assign Web
traffic to the bronze priority setting.

Managing Class-of-Service Thresholds

“XO Applications Once the applications on the network have been identified, configured, and categorized
Performance Management correctly, it is critical not to exceed the bandwidth thresholds allocated by the service
provider. Because of the bursty nature of data traffic, The enterprise might be well
monitors utilization for each
below its allocation for the vast majority of the day, but at peak times, it exceeds the
individual CoS and measures threshold. Consequently, mission-critical applications might be negatively impacted due to
whether usage is above or reclassification or discarding.
below threshold allocations.”

Figure 6: Determine if your actual usage is exceeding your

policing threshold

Tools, such as XO Applications Performance Management, monitor utilization for each

individual CoS and measures whether usage is above or below threshold allocations. If you
are above your threshold allocation, you either need to upgrade your circuit for additional
bandwidth or move one or more applications to another CoS (see Figure 6). While there are
additional considerations for managing CoS thresholds, they can be addressed easily with
APM.

Connect / Communicate / Manage / Protect / Optimize 13

The Benefits of Migrating to MPLS

Measuring Service Level Agreements

XO Applications Performance Management is a leader in SLA verification by providing

independent analysis and measurements from an end-to-end perspective. Instead of waiting
for a monthly report from your provider, you receive an up-to-the-minute view of service level
parameters, including availability, delay, and throughput. The up-to-the-minute view is critical
in isolating poor performance between the network and other aspects. For example, you can
quickly determine if a long, round-trip delay is the cause of the application degradation.

Optimizing Bandwidth Requirements

The cost of bandwidth alone can comprise as much as 60 to 70 percent of total networking
budgets for some enterprises. With tighter resources, optimizing bandwidth is a key step
toward maximizing network and application performance. With the new Layer 3 connectivity—
IP addressing—instead of Layer 2 connectivity—PVCs— utilization will change in a private IP
environment.

Adding CoS capabilities also throws a new wrinkle into the equation. With the bursty nature
of traffic, it is important to understand the impact of usage on a private IP environment. The
Burst Advisor in XO Applications Performance Management measures actual usages in one-
second increments in order to size your circuits properly. Consequently, the networking group
would have the information to make decisions, such as increasing bandwidth for critical
locations or moving bandwidth from less-utilized sites to over-utilized sites.

Managing the Network Migration

“With tighter resources,
During a private IP implementation, there will more than likely be a period of time where
optimizing bandwidth is a multiple technologies—such as frame relay, ATM and private IP—are being run over the
key step toward maximizing network. With a mixed network, it is mandatory to maintain visibility into the performance
of the network and applications. Ideally, enterprises would baseline performance based on
network and application
existing technology and, as the migration progresses, IT staff would see the changes and
performance.” impact of Layer 3 connectivity and CoS.

XO Applications Performance Management offers a unique opportunity to manage the

transition to private IP. By managing frame relay, ATM, and private IP, enterprises equip the
network with frame or ATM as a baseline and have a single platform to manage network
application integrity as they migrate to private IP. XO Applications Performance Management
provides visibility into interworked environments so you can still see end-to-end performance,
even if one site is frame relay and the other is private IP (see Figure 8). Now you have
the ability to monitor either PVCs in a frame relay environment or the harder challenge of
managing IP-subnet pairs in a private IP network.

14 Solutions Built Around You.

 XO Communications

“XO Applications
Performance Management
provides visibility into
interworked environments so
you can still see end-to-end
performance, even if one site
is frame relay and the other is
private IP.”

Figure 8: End-to-end network visibility

Conclusion

As you begin or continue your consideration of MPLS-based/private IP networks for your

enterprise, it is extremely important to consider the benefits of migration as well as what
will change as you migrate. There are many key reasons or benefits for most enterprises to
implement a private IP network, which why a growing number of enterprises have already
implemented an MPLS-based network.

MPLS networks are no longer on the edge of innovation due to their growing adoption.
However, there are still key hurdles that must be addressed, ranging from CoS prioritization
to understanding the change from PVC-based connectivity to IP subnet-based connectivity.
The good news is there are tools available to bridge the challenge of migration to a private IP
network.

Connect / Communicate / Manage / Protect / Optimize 15

The Benefits of Migrating to MPLS

About XO MPLS & Application Performance Management

MPLS-based private IP networks provide many benefits for enterprises, ranging from increased
performance to less complexity to lower costs, but this also changes the infrastructure IT
managers have administered for years. XO Applications Performance Management leverages its
history in traditional network performance management with enhanced private IP and application
functionality so IT managers can confidently make the decision to migrate to an MPLS-based
network if it benefits the enterprise.

The XO Applications Performance Management tool integrates the analysis of application

performance with analysis of network performance – so IT departments can take a holistic
approach to network application integrity. Now enterprises can understand what applications
are on the network so they can be assigned correctly, as well as monitor and manage the more
complex IP subnetto- IP subnet connectivity between each and every location in the network.

About XO Communications
XO is a leading provider of 21st century communications services for businesses and
communications services providers, including 50 percent of the Fortune 500 and leading cable
companies, carriers, content providers and mobile operators. Utilizing its unique and powerful
nationwide IP network, extensive local metro networks and broadband wireless facilities,
XOH offers customers a broad range of managed voice, data and IP services in more than 80
metropolitan markets across the United States.

For additional information, please visit XO Communications at www.xo.com or call directly at

888-718-7595 for more information.

XOMPLSWP-1110

© Copyright 2010. XO Communications, LLC. All rights reserved. 16

XO, the XO design logo, and all related marks are registered trademarks of XO Communications, LLC.