Transcript MPLS

MPLS (MultiProtocol Labeling Switching)
School of Electronics and Information
Kyung Hee University.
Choong Seon HONG
<[email protected]>
Introduction




Multi-protocol Label Switching
A switching and forwarding scheme
Evolved from Cisco’s Tag Switching
It located between Layer 2 and Layer 3 of
the OSI reference model
 It make use of the fixed length label (20-bit)
for switching and forwarding
2
Introduction
 Major components:
 LERs (Label Edge Routers)
• It located at the boundary of the MPLS network and its
function is assignment and removal of labels as packet
enter end leave the MPSL network respectively.
 LSRs (Label Switching Routers)
• It located at the core part of the MPLS network and its
function perform packet switching based on the label.
3
Introduction
 Major components (cont.)
 LDP (Label Distribution Protocol)
• Maps unicast IP address into MPLS labels
 LSPs (Label-switched Paths)
• A flow of MPLS packets with same label
• Similar to VC (Virtual Circuit) in ATM network
4
Introduction
 General operations:






label creation and distribution
table creation at each router
label-switched path creation
label insertion/table lookup
packet forwarding/switching
Label removal
5
Introduction
 Applications:
 Used in core network to improve packet
forwarding performance
 Support QoS and CoS application
 Improve network scalability (use of LSPs)
 Integrated IP and ATM network
 IP-VPN
6
What is Multi-protocol Label Switching ?
MPLS Technology
 Routers or switches that handle MPLS and IP are
known as Label Switch Routers (LSR’s)
 LSR’s at the edge of MPLS networks are sometimes
referred to as Label Edge Routers (LER’s)
 Ingress LER’s are responsible for classifying unlabelled IP packets and
appending the appropriate label.
 Egress LER’s are responsible for removing the label and forwarding the
unlabelled IP packet towards its destination.
 All IP packets that follow the same path through the MPLS network and
receive the same treatment at each node are known as a Forwarding
Equivalence Class (FEC).
8
Label Switching Devices
Label Switching Routers (LSRs)
(ATM Switch or Router)
Label Edge Routers
9
MPLS Technology
There are three key elements of MPLS
 The MPLS header stack
• Which contain the MPLS label on which Label Switch Routers
will forward the packet. Headers can be stacked.
 The enhanced IP routing protocols
• Which distribute topology and constraint based data
 The label distribution protocols
• The standardized connection establishment protocols through
which LSR’s set up a complete path from ingress LSR to egress
LSR. This path is known as a Label Switched Path or LSP.
MPLS adds a connection oriented paradigm into IP networks
10
MPLS Shim Header Structure
MPLS Headers can be used “recursively”
MPLS "shim" headers
...
Layer 2 Header
Label
Exp. S
4 Octets
Label Switching
Look up inbound label + port (+Exp)
to determine
outbound label + port + treatment
TTL
IP Packet
Label:
Exp.:
S:
TTL:
20-bit value, (0-16 reserved)
3-bits Experimental (ToS)
1-bit Bottom of stack
: Stack Indicator
8-bits Time To Live
Header operations
Swap (label)
Push (a new header)
Pop (a header from stack)
MPLS encapsulations are also defined for ATM and Frame relay.
11
Establishing the label bindings
 Each switch needs a table that contains the actions it is to
perform when a given label arrives.
 The downstream end of a link needs to know what label values
will be sent.
 This can be done by management action – directly equivalent to
PVC’s in ATM.
 But this does not scale well.
 And there is no interoperability between management systems –
so multi-operator connections are difficult if not impossible.
 Hence trend to protocol driven service establishment and the
reason for IP’s success.
 So we need to automate the LSP establishment process.
12
MPLS Routing protocols
Start with existing IGP’s
OSPF
IS-IS
BGP-4
Distribute topology
information only
 Enhance to carry constraint data
OSPF-TE
IS-IS –TE
Constraint data
Link capacity,Link utilization
Resource class
Priority
Pre-emption etc
Constraint based routing is the key to Traffic Engineering
13
Label Distribution Protocols
 LDP
 CR-LDP
 RSVP-TE
Hop by Hop routing
Ensures routers agree on bindings between
FEC’s and the labels.
Label paths follow same route as
conventional routed path
Explicit constraint based routing
Route determined by ingress LSR based
on overall view of topology, and constraints
Traffic engineering
CoS and (QoS)
fast (50ms) rerouting
14
MPLS Partitions Routing and Forwarding (1)
Based on:
Classful Addr. Prefix?
Classless Addr. Prefix?
Multicast Addr.?
Port No.?
ToS Field?
Routing
OSPF, IS-IS, BGP, RIP
Forwarding Table
Forwarding
Based on:
MPLS
By separating Routing from forwarding
MPLS introduces more flexibility to develop
new routing solutions without impacting the
data plane hardware of label switch routers
Single forwarding paradigm – multiple
routing paradigms
Exact Match on Fixed Length Label
The edge LSR is able to use
a wide variety of input in
determining the FEC, and not
just the destination IP
address
Flexibility in forming FEC’s
15
MPLS Partitions Routing and Forwarding(2)
 Combines Layer 3 routing with
label-swapping forwarding
 Simplicity of Layer 2 forwarding offers high performance
 Layer 3 routing has proven scalability
 Clean separation of Forwarding and Control/Routing
 Forwarding component: Simple label-swapping paradigm
 Control component: Collection of modules to maintain and
distribute label bindings
 Separation leads to graceful evolution of control paradigm
16
Forwarding Component
Label Forwarding Information
Base (LFIB)
Each entry consists of:
• Incoming label
• One or more sub-entries:
–Outgoing label, outgoing interface, outgoing MAC address
LFIB is indexed by incoming label
17
Forwarding Component (Cont.)
Forwarding algorithm:
Extract label from a packet
Find LFIB entry with
incoming label = label from packet
Replace label in packet with outgoing label(s)
Send packet on outgoing interface(s)
Observation: forwarding algorithm is
Network Layer-independent
independent of how labels have been assigned (ie
by Control module)
18
Label Switching Example
Destination-Based Routing Module
Address
Prefix Interface
Address
Prefix Interface
128.89.10
1
128.89.10
0
171.69
1
171.69
1
...
128.89.10
...
i/f 0
i/f 1
i/f 1
Advertises Reachability
to 128.89.10
Advertises Reachability
to 128.89.10 and 171.69
171.69
Advertises Reachability
to 171.69
19
Label Switching Example (Cont.)
Address
Prefix Interface
Address
Prefix Interface
128.89.10
1
128.89.10
0
171.69
1
171.69
1
...
128.89.10
...
i/f 0
i/f 1
Advertises Binding
<5,128.89.10> Using LDP
i/f 1
Advertises Bindings
<3,128.89.10>
<4,171.69> Using LDP
171.69
Advertises Binding
<7,171.69> Using LDP
20
Label Switching Example (Cont.)
Local Remote Address
Label Label
Prefix
Interface
x
3
128.89.10
1
x
4
171.69
1
Local Remote Address
Label Label
Prefix
Interface
3
5
128.89.10
0
4
7
171.69
1
...
128.89.10
...
0
1
171.69.12.1 data
1
4
7
171.69.12.1 data
171.69.12.1 data
‘Edge’ Router Does
Longest Match, Adds Label
Subsequent Routers
Forward on Label Only
171.69
21
Label Distribution for ATM Downstream on Demand
Requests a label
for 128.89
Input Local Remote Address Output
i/f
Prefix
i/f Label Label
1
5
7
128.89
0
2
6
8
128.89
0
...
128.89
Requests a label
for 128.89
Requests Two Labels
for 128.89
Returns a Label to
Each Requester
Label Switching = ATM switching
because labels copied in VCI
How does it fit into IP network development plans
–
MPLS Applications
Applications of MPLS
 Traffic Engineering
 Adding Class of Service (CoS) and Quality of
Service (QoS)
 Network scalability
 Supporting IP VPN’s
24
Traffic Engineering
Current IGP’s lead to Hyper-Aggregation
TRAFFIC FOR D
SHORTEST PATH ROUTED
D
S
CONGESTION
MASSIVE
CONGESTION
25
Traffic Engineering
Current IGP’s lead to Hyper-Aggregation
TRAFFIC FOR D
SHORTEST PATH ROUTED
9 UNDER ULTILIZED]
LINKS
3 OVERUTILIZED
]
D
S
CONGESTION
MASSIVE
CONGESTION
26
Traffic Engineering IS the Answer
 Objectives
Map actual traffic efficiently to available resources
Controlled use of resources
Redistribute traffic rapidly and effectively in
response to changes in network topology particularly as a consequence of line or equipment
failure
 Note this complements Network Engineering
Putting the network where the traffic is
27
Traffic engineering distributes traffic
Traffic distributed over
Network resources by
MPLS traffic engineering
- Congestion eliminated
D
S
28
Adding CoS and QoS
 Explicit path set up can also associate
specific resource requests with an FEC
 Class of service
Establish relative priority of one FEC over another
– no absolute guarantees
 Quality of service
Specific guarantees on
• Bandwidth
• Delay
• Burst size etc
CoS and QoS require
explicit support in the
data plane of the LSR’s
 Primary objective is for MPLS to support the
Diff-Serv QoS model (EF, AF1-12,etc)
29
Hierarchy via Label stack= Network scalab
ility
Layer 2 Header
Within each domain
the IGP simply needs
to allow the Boarder
(ingress) routers to
determine the
appropriate egress
boarder router
Reducing drastically
size of routing table in
transit routers
Label 3
Label 2
Label 1
IP Packet
MPLS Domain 1
MPLS Domain 2
MPLS Domain 3
30
Benefit of MPLS in scaling
 MPLS labels introduce hierarchy
 Transit routers no longer need to handle
complete routing tables
 New layers of the hierarchy can be
introduced as needed for scaling.
31
Support of IP VPN’s
 A Virtual Private Network
An IP network delivering private network services
over a public infrastructure
Supports global and non unique private address
space
Supports CoS and QoS
 Use of labels isolates IP addresses within
public network from customer IP addresses
 Creates a highly scalable VPN
32
Benefit of MPLS IP VPN’s
 Provides a mechanism to scale both the number of
VPN’s and the number of members per VPN to very
large numbers.
 Allows VPN’s to have non-unique IP addressing
 Provides for a great deal of flexibility in defining the
VPN service (from the mapping to FEC’s)
 Enables meaningful CoS and QoS Service Level
Agreements (SLA’s) to be associated with a VPN
33
Why MPLS VPNs?
MPLS combines L3 routing and L2 forwarding
L3 routing provides
improved scalability by eliminating mesh of
connections from CPE-to-CPE
L2 (label-based) forwarding provides
comparable security to L2 approaches
hiding of non-registered addresses
Hierarchical labels (label stack) further
enhance scalability
34