MPLS
Traffic Engineering
MPLS
Traffic Engineering
George Swallow
[email protected]
George Swallow
[email protected]
What is Traffic Engineering
What is Traffic Engineering
Taking control of how traffic flows in your network in order to
-ÿImprove overall network performance ÿOffer premium services
ÿAs a tactical tool to deal with network design issues when the longer range solution are not deployed
Voice Traffic Engineering
•
Telco’s noticed that demands vary
widely by time of day
•
Began “engineering the traffic”
long ago
•
Evolved over time
Reasons for Traffic Engineering
•
Economics – more packets, fewer $$$
•
Address deficiencies of IP routing
•
Tactical tool for network operations
•
Class-of-service routing
Mike O’Dell, UUnet
Economics of Traffic Engineering
“
“
The efficacy with which one uses the
The efficacy with which one uses the
available bandwidth in the transmission
available bandwidth in the transmission
fabric directly drives the fundamental
fabric directly drives the fundamental
‘
‘
manufacturing efficiency
manufacturing efficiency
’
’
of the
of the
business and its cost structure.
business and its cost structure.
”
”
Savings can be dramatic. Studies have shown that transmission costs can be reduced by 40%.
The “Fish” Problem
a deficiency in IP routing
The “Fish” Problem
a deficiency in IP routing
IP uses shortest path destination based routing
IP uses shortest path destination based routing
Shortest path may not be the only path
Shortest path may not be the only path
Alternate paths may be under-utilized while the
Alternate paths may be under-utilized while the
shortest path is over-utilized
shortest path is over-utilized
R8
R2
R6
R3
R4
R7
R5
Deficiencies in IP Routing
Deficiencies in IP Routing
• Chronic local congestion
• Load balancing
Across long haul links
• Size of links
Difficult to get IP to make good use unequal size links without overloading the lower speed link
Overlay Motivation
Separate Layer 2 Network
(Frame Relay or ATM)
Mike O’Dell UUnet, November 17, 1996 “
“The use of the explicit Layer 2 transit layerThe use of the explicit Layer 2 transit layer gives us very exacting control of how
gives us very exacting control of how
traffic uses the available bandwidth in
traffic uses the available bandwidth in
ways not currently possible by tinkering
ways not currently possible by tinkering
with Layer 3-only metrics.
The Overlay Solution
•
Layer 2 network used to manage
the bandwidth
•
Layer 3 sees a complete mesh
L3 L3 L3 L3 L3 L3 L3 L3 L3 L3 L3 L3 L3 L3 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L3 L3 L3 L3 L3 L3 L3L3 L3L3
Overlay Drawbacks
•
Extra network devices (cost)
•
More complex network management
Two-level network without integrated NM Additional training, technical support,
field engineering
•
IGP routing doesn’t scale for meshes
Number of LSPs generated for a failed router is O(n3); n = number of routers
Traffic Engineering & MPLS
• MPLS fuses Layer 2 and Layer 3
• Layer 2 capabilities of MPLS can
be exploited for IP traffic engineering
• Single box / network solution
+
=
ororRouter ATM Switch MPLS
An LSP Tunnel
An LSP Tunnel
R8
R2
R6
R3
R4
R7
R5
R1
Normal Route R1->R2->R3->R4->R5 Tunnel: R1->R2->R6->R7->R4
Labels, like VCIs can be used to establish virtual circuits
LSP Tunnel Setup
LSP Tunnel Setup
Setup: Path (R1->R2->R6->R7->R4->R9) Tunnel ID 5, Path ID 1
Setup: Path (R1->R2->R6->R7->R4->R9) Tunnel ID 5, Path ID 1
22 49 17 R8 R2 R6 R3 R4 R7 R1 R5 R9
Reply: Communicates Labels and Label Operations
Reply: Communicates Labels and Label Operations
Reserves bandwidth on each link
Reserves bandwidth on each link
Pop
Multiple Parallel Tunnels
•
Automatically load shared
•
Weighted by bandwidth
to nearest part in 16
•
Traffic assigned by either
Source-Destination hash Round robin
Automatic Load Balancing
Automatic Load Balancing
New York #1 New York #2 Washington London #1 London #2 Paris Stockholm Munich Brussels Frankfurt Amsterdam Link #1 Link #2 Link #3 LSP Tunnel #1
LSP Tunnel #2
Additional Features
Additional Features
•
Adjusting to failures
Requires rapid notification
•
Adjusting to improvements
•
Need to account for
Global optimality Network stability
Protection Strategy
Protection Strategy
Two pronged approach:
• Local protection
Repair made at the point of failure us to keep critical applications going
Fast - O(milliseconds) Sub-optimal
• Path protection
An optimized long term repair Slower - O(seconds)
Local Protection via a
Bypass Tunnel
Local Protection via a
Bypass Tunnel
R2
R3
R6
R4
R8
R7 R1
R10 R9
R5
Primary Paths Bypass Tunnel
Path Protection
Path Protection
R2
R3
R6
R4
R8
R7 R1
R10 R9
R5
Primary Path Backup Path
Summary
Summary
Traffic engineering provides the means to
ÿSave transmission costs
ÿAddress routing deficiencies
ÿAttack tactical network engineering problems
ÿProvide better QoS
Making sure resource are available Minimizing disruption
