Network Virtualization

4/30/2012
1
Group 8











Trung Kin
V Hng Vit
H Vn Sng
Nguyn Tin S
Phm Vit Hng
T Thanh Tng
Hong V Thnh Lun





Contents
2
1. Historical perspective
2. Introduction about Network Virtualization (NV)
3. Network virtualization: concepts and
challenges
4. iMark
5. Cross layer interaction
6. ViNE- Yard

1. Historical perspective
3
1.1. Virtual Local Area Network
1.2. Virtual Private Network
1.3. Active and Programmable Networks Overlay
Networks

1.1. Virtual Local Area Network
4
A VLAN is a group of logically networked hosts
with a single broadcast domain regardless of
their physical connectivity.
1.2.Virtual private network
5
A VPN is a dedicated network connecting
multiple sites using private and secured
tunnels over shared or public communication
networks like the Internet.

1.3. Active and Programmable Networks
Overlay Networks
6
Active and Programmable Networks
Customized network functionalities
Programmable interfaces and active
codes
Overlay Networks
Application layer virtual networks
Not flexible enough
2. Introduction about NV

7
Why Network Virtualization?
Internet is almost ossified
Lots of makeshift solutions(e.g overlay)
Hard to come up with onesizetall architectu
re
Almost impossible to predict what future mig
ht unlash
Why not create allsizestintoone architectu
re instead
Open and expandable
Coexistence of heterogeneous architecture
2. Introduction about NV

8
What is Network Virtualization?
Transparent abstraction of network platform
and resources
Multiple virtual networks(VNs)
Additional level of indirection
Indirect access to network resources
Resource partitioning and isolation
Physical and logical
2. Introduction about NV

9
Virtualized everything
Virtual machine: Vmware, Xen
Storage virtualization: SAN
Data-center virtualization
Network virtualization is an architectural
approach to providing a separate logical
networking environment for each group within
the enterprise.
These logical environments are created
over a single shared network
infrastructure
2. Introduction about NV

10
Making a physical network appear as multiple
logical ones

2. Introduction about NV

11
Network Virtualization Environment
2. Introduction about NV
12
What is virtual
network ?

2. Introduction about NV

13
What is virtual network ?
A collection of virtual nodes and virtual links
forming virtual topology
Subset a physical topology
Basic entity of NVE
A virtual node is hosted on a particular
physical node
Multiple virtual nodes can coexist
A virtual link spans over a physical path
Includes a portion of the underlying physical
resources

2. Introduction about NV

14
Business Model

2. Introduction about NV

15
Business Model

2. Introduction about NV

16
Design Principles

2. Introduction about NV

17
Design Principles
Concurrence of heterogeneous virtual
networks
Introduces diversity
Recursion of virtual networks
Opens the door for network virtualization
economics
Inheritance of architectural attributes
Promotes value-addition
Revisitation of virtual nodes
Simplifies network operation and management

2. Introduction about NV: on going
Project
18
3. NV: Concepts and challenges

19
3.1. Concepts
3.2. Challenges
3.3. Future Works
3.1. Concepts

20
What can network virtualization deliver?
Coexistence of novel heterogeneous networking
technologies with legacy networks
Cost efficiency due to improved resource
utilization
Sharing in isolation
Improved Manageability
Clear separation of policy from mechanism
Defined accountability of infrastructure and service
providers
Modular management

3.1. Concepts

21
What can network virtualization deliver?
Scalability
Maximize the number of co-existing virtual networks
Increase resource utilization and amortize CAPEX and
OPEX
Security, Privacy, and Isolation
Complete isolation between virtual networks
Logical and resource
Isolate faults, bugs, and misconfigurations
Secured and private
Programmability
Of network elements e.g. routers


3.1. Concepts

22
Legacy Support
Consider the existing internet as a member of the
collection of multiple virtual internets
Very important to keep all concerned parties satisfied
Increasing network availability by quarantining
hosts that are infected by viruses or not compliant
with the enterprise security policies.
Creation of secure network areas that are partially
or totally isolated.
Consolidation of multiple networks onto a single
infrastructure.

3.2. Challenges

23
3.2. Challenges

24
Management problems to be Addressed

3.2. Challenges

25
3.3. Future Works

26
Future Works
Instantiation : ViNe-Yard
Creates virtual networks
Logistics
Runs them
Management : iMARK
Manages them
Interactions
Let them flourish

3.3. Future Works

27
Enabling technologies
In this section, we provide an overview of the
varios link and node virtualization technologies
that exist today and could be used to form a
virtualized network
VLANs (Virtual Local Area Networks) : allowing the
creation of isolated networks where all participating
nodes are in a single broadcast domain
VPNs (Virtual Private networks): are created by using
MPLS(Multi-protocol Label Switching)
Node virtualization aims to provide efficient
sharing and isolation of computing resources

3.3. Future Works

28
A Virtualization Technologies Primer: Theory
Device
How to traffic separation maintained internally to a
device?
Data path
How is traffic separation enforced across a network
path
Control plane
Because data-path virtualization essentially builds an
overlay topology, what changes are needed for routing
protocols to function correctly?

3.3. Future Works

29
Device virtualization
Depending on the type of device, the virtual
separation can go by the following names:
Virtual LAN(VLAN)
Switch MAC Table
Virtual routing and forwarding (VRF)
Virtual forwarding instance (VFI)
Virtual firewall context
Data-path Virtualization
Connecting devices with private paths over a
shared infrastructure is a well-known problem
Label Switched Paths
3.3. Future Works

30
Data-path Virtualization
Connecting devices with private paths over a
shared infrastructure is a well-known problem
Label Switched Paths

3.3. Future Works

31
Control-Plane Virtualization Routing Protocols
The most familiar virtualized control plane is
probably Per VLAN Spanning Tree (PVST), which
has a separate spanning-tree instance for each
VLAN running on a switch

4. iMARK
32
Motivation
Architecture Overview
iMark Operation


An identity management framework in the
Network Virtualization Environment
4.1.Motivation
33
Dynamism in the NVE
- Macro Level: Merge/Separate VNs, to creat
compound VNs, this is known as federation of
VNs. Multiple federations and VNs can also
come together to creat hierarchy of VNs
- Micro Level: Dynamic join, leave, and
mobility of the end users within and in between
multiples VNs.
Dynamism incurred by the migration of virtual
routers.

4.1.Motivation
34 Scale: the number of users in increasing
rapidly, result as naming infrastructure for
NVE of something else must be scalable.
Interactions between Multiple heterogeneous
parties: the way multiple players among
themselves. Such an interaction between two
SPs, or two InPs, or an SP and an InP,

4.1.Motivation
35
Uber-homing:
- In NVE, end users can simultaneously
connect to multiple VNs through multiple InPs
using heterogeneous technologies to acess
different services.
- Significant impact in cross VN routing.


4.2. Architecture Overview

36
Design principle
- Separation of Identity and Location:
Supporting logical and physical mobility,
federation, and hierarchy of VNs, and uber-
homing.
- Local Autonomy: Defined interfaces and
mechanisms to enable end to end connectivity
across heterogeneous physical and virtual
networks.
Flexibility of naming and addressing in
different VNs.
- Global Identifier Space: Provide
communication between nodes in different
VNs a globally agreed upon identification
mechanism.


4.2. Architecture Overview
iMark component and concept


37
Controllers:
adress
allocation,
name
resolution,
Adapters: as
gateways
between
VNs
4.2. Architecture Overview
iMark component and concept


38
Entities and identifier spaces
4.2. Architecture Overview
iMark component and concept


39
Mappings: between different identifiers
4.2. Architecture Overview
iMark component and concept


40
Federation and Hierarchy of iMark Controllers
4.3. iMark Operations
41
Macro level Operations

Creat VNs

Federation


Hierarchy
4.3. iMark Operations
42
Micro level
- Join:
Connect to a
VN

4.3. iMark Operations
43
Micro level
- Lookup and
Connection
Setup: State
setup in the
network
4.3. iMark Operations
44
Micro level
- Leave: an end host wants to leave a VN
- Mobility:
+ Geographical mobility: soft handoff
+ Logical mobility: Leave and then Join
4.4. Sumary
45
iMark manages identifiers for entities at different
level
- Macro level: assists creation of independent
VNs and formation of VN federations and
hierarchy of VNs.
- Micro level: enables end to end communication
between end hosts in different VNs.
Separate identity of the end hosts from their
physical and logical location, with help of a global
identifier space, provide universal connectivity
without revoking the autonomy of the concerned
physical and virtual networks
6. Analyzing Crosslayer Interaction in
Overlay Networks

Analyzing Crosslayer Interaction
in Overlay Networks
Overlay Networks
Overlay networking helps
overcome functionality
limitations of the Internet
by forming a virtual network
over the native IP network

Classification of Overlay Networks
Overlay networks
Peer-to-peer
networks
(e.g. BitTorrent)
End-system
overlays
(e.g. Skype)
Routing overlay
networks
Service
overlays
(e.g. VINI)
Service Overlay Networks (contd.)
Many types of services can be offered
Multicast (e.g. ESM, Overcast)
QoS (e.g. OverQoS, SON)
Security (e.g. DynaBone, SOS)
Better routes (e.g. RON, Detour, X-Bone)
and much more
Service Overlay Networks
Offer enhanced or new services by deploying
intelligent routing schemes.

Relaying
Overlay link
Implementation
Typically, we just add a new header inside the
regular IP header






SRC ---> encapsulator --------> decapsulator ---> DEST

Referred to as IP-in-IP encapsulation ( IP tunneling)
Ethernet
header
IP
header
TCP
header
Application
data
Secondary
IP header
WIRE
Dest B Dest C
Cross-Layer Interaction
Performing dynamic routing at both overlay and
native IP layers leads to:

Conflict due to mismatch or misalignment of
routing objectives
Contention for limited physical resources
Functionality overlap (Both overlay layer and IP
layer perform similar set of functions)
Dual Rerouting
Each layer performs rerouting, with no
knowledge of which layer leads to optimal
restoration
A
OVERLAY
1
LAYER
NATIVE IP
LAYER

X
A
D
E F
H
F
H
G A
E
C
C
B
D
B
G
Overlay
rerouting
Native rerouting
Failure
Solutions for Improved Recovery
Probabilistically suppressed overlay rerouting

Deferred Overlay rerouting

Follow-on Suppressed overlay rerouting


Tuning Dual Rerouting
Intra-domain
(keepAlive-time = 1 sec, hold-time = 3 secs)
Dual
Rerouting
Suppress
overlay
rerouting at
0.5 prob.
Defer overlay
rerouting by
0.375 secs
Native-only
rerouting
Average route
changes 125.08% 101.59% 109.85% 1.567
Stabilized
inflation 100% 108.32% 100% 1.202
Time when
stable 113.7% 100.48% 107.33% 2.481
Peak inflation 114.22% 109.98% 110.73% 1.202
Further Improving Recovery
Adjust the functioning of native layer:
Tuning the native layer keepAlive-time:


This produces the best
tradeoff between # of
route changes,
stabilization time and
recovery time
k
e
e
p
A
l
i
v
e
-
t
i
m
e

k
e
e
p
A
l
i
v
e
-
t
i
m
e

Tuning
FUTURE WORK
OVERLAY-FRIENDLY NATIVE
NETWORKS
ViNE_Yard
(Virtual Network Embedding with
Coordinated Node and Link Mapping)
6. ViNE_Yard
What is VN Embedding?
Given:
Single substrate network: G
s
=(N
s
,E
s
)
Online VN request : G
V
=(N
V
,E
V
)
Requirements and Constraints of virtual nodes and
virtual links.
Task :
Assign virtual nodes and links to substrate nodes and
links.
Allocate resources
CPU, bandwidth
VN Embedding Objectives
Maximize
Acceptance ratio
Percentage of request accepted
Revenue
Based on resources requested for a VN
Minimize
Cost
Based on substrate network resource allocated for
embedding VN request.
ViNE Algorithm
D_ViNE (Deterministic Virtual Network Embedding)
R_ViNE (Random Virtual Network)

Virtual Network Embedding problem
1. Substrate Network : G
s
=(N
s
,E
s
)
N
s
: the set of substrate nodes
E
s :
the set of substrate links
n
s
N
s
:substrate node
c(n
s
) : CPU capacity
loc(n
s
): geographic location of n
s
.
e
s
(i,j) E
s
: substrate link between node i and j
b(e
s
): total amount of bandwidth.
P
s
(s,t) : substrate path from source node s to destination node
t.



Virtual Network Embedding problem
Virtual Network Embedding problem
2. Virtual Network Request
G
v
=(N
v
,E
v
)
N
v
: the set of virtual node
E
v
: the set of

virtual link
D
v
> 0 :expressing how far a virtual node n
v
N
v
can be placed
from the location specified by loc(n
v
).

Virtual Network Embedding problem
3. Residual Capacity of substrate Resources
A substrate node stress
S
N
(n
s
) : the total amount of CPU capacity allocated to virtual
nodes hosted on the substrate node.



A substrate link stress
S
E
(e
s
) : the total amount of bandwidth reserved for the virtual
links


Virtual Network Embedding problem
Residual of substrate node:


Residual of substrate link


Total amount of bandwidth on substrate path

Virtual Network Embedding problem
4. Virtual Network Assignment
Node assignment





Subject to

Virtual Network Embedding
problem
Link assignment



Subject to

Virtual Network Embedding
problem
4. MIP formulation for optimal Embedding VN.
4.1. Substrate Graph Augmentation

Virtual Network Embedding
problem
4.2 Relaxed Linear Program for ViNE
Variable
A flow variable denoting the total amount of
the flow in the on the substrate edge(u,v) for
the ith virtual edge
A binary variable, which has the value =1 if
else =0


i
uv
f
u v
uv
i
f
i
uv
x
( ) 0
i i
i uv vu
f f + >
Virtual Network Embedding
problem
Objective






Parameter to control the importance of load balancing while
embedding request.

w
1 ( , )
1 (w)
uv E
N
R u v
R
o
|
< <
< <
Virtual Network Embedding
problem
Contraints
Capacity Contraints:




The node and edge capacity bound
To ensure summation of flows on both direction remain
within available bandwidth.



Virtual Network Embedding problem
Flow Related Contraints:






Virtual Network Embedding
problem
Meta and Binary Constraints:
Related to the augmented portition of
substrate graph
Make sure that only one substrate node
is selected for each meta-node.








Domain Constraints:

ViNE-Yard(D-ViNE&R-ViNE)
D-ViNE:Deterministic Rounding-based Virtual
Network Embedding Algorithm
D-ViNE:Deterministic Rounding-based Virtual
Network Embedding Algorithm
Creating an Augmented substrate graph G
S
=(N
S
,E
S
)
Solve Relaxed Linear program to get fractional solution
Checking where there are any unmapped substrate nodes within its
feasible region.
Calculating P
z

D-ViNE maps the virtual node onto the unmapped substrate node z
with the highest value P
z
,
D-ViNE applies MCF ( multi-comodity flow) algorithm to map virtual
edge
Update the residual capacity




max
( ) 1 z
R-ViNE:Deterministic Rounding-based Virtual
Network Embedding Algorithm
Performance Evaluation
1. Simulation Setting
Substrate network
50 nodes in a 25x25 grid with 0.5 link probability
CPU/BW uniformly distributed in the range: 50-100
units
VN requests
Poisson arrival rates from 4 VN request 100 times
units
Exponentially distributed lifetime of 1000 time units
2-10 nodes with 0.5 link probability
Tools: GT-ITM ,GLPK
Performance Evaluation
2. Performance Metrics
2.1. Acceptance Ratio
2.2. Generated Revenue
2.3. Provisioning Cost
2.4. Average Node Utilization
2.5. Average Link Utilization
Acceptance Ratio

Generated Revenue

Average Node Utilization
Average Link Utilization



THANK YOU FOR YOUR ATTENTION
COMMENTs AND QUESTIONs