Traffic Engineering

within MPLS
Information Distribution
Sources:
MPLS Forum
E. Osborne and A. Simha, Traffic Engineering with MPLS, Cisco Press

Slide 1
MPLS Traffic Engineering –
Information Distribution
• Value added services enabled by MPLS Traffic Engineering
 Constraint-based routing
 QoS
 Fast reroute
 VPNs
 …

• What’s involved in information distribution to support TE

Slide 2
Review Terminology...

• Network Engineering
 "Put the bandwidth where the traffic is"
 Physical cable deployment
 Virtual connection provisioning

• Traffic Engineering
 "Put the traffic where the bandwidth is"
 Local or global control
 On-line or off-line optimization of routes
 Implies the ability to “explicitly” route traffic

Slide 3
 Traditional Traffic Engineering

C1

C3

C2

Layer 3 Routing Traffic Engineering

• Move traffic from IGP path to less congested path

Slide 4
Traditional Traffic Engineering
Limitations

C1

C3

C2

• TE Mechanisms • Limitations
 Over-provisioning  Some links become
underutilized or overutilized
 Metric manipulation
 Trial-and-error approach
Slide 5
 Traffic Engineering with ATM Core

Virtual Circuit

Physical
Topology

• Infrastructure
 Routed edge over ATM switched core
 Introduced full Traffic Engineering (TE) ability

Slide 6
Traffic Engineering with ATM Core
Limitations

Logical
Topology

• TE Mechanisms • Limitations
 VC routing  Overlay of IP and ATM
 Overlay network  “N-squared” VCs
• Benefits  IGP Stress
 Full traffic control  Cell tax
 Per-circuit statistics

Slide 7
MPLS Traffic Engineering

• Traditional TE controls traffic flows in a network

 “The ability to move traffic away from the shortest path
calculated by the IGP to a less congested path”
• MPLS Traffic Engineering
 Allows Explicit Routing and set-up of LSP’s
 Provides control over how LSP’s are recovered in the
event of a failure
 Enables Value Added Services
 Virtual Private Networks – VPNs
 Service Level Agreements - SLAs
 Multi-media over IP solutions – MMoIP, VoIP
 ATM over IP – easy andSlide
cheap
8 for existing legacy networks
 Review Topology and LSP set-up

LSRs use routing protocols to IP networks advertise their

discover network topology addresses using routing
protocols into the MPLS
eg. OSPF, ISIS, (BGP)
cloud

Network Router1 Router2 Network

A B

LSR2 LSR3

LSR1 LSR4
LSR5
An LSP is set up
between these networks LSR6

MPLS Domain

Slide 9
But, this is a simple example . . .
• Routing Protocols Create a "Shortest Path“ Route
• LSPs follow the "shortest path"
C1

C3

C2

This mechanism does NOT give us

Traffic Engineering

Slide 10
MPLS Traffic Engineering
Requires 3 main areas of extensions
• Enhancements to the Routing Protocols: Information
Distribution
 OSPF  OSPF-TE
 ISIS  ISIS-TE

• Enhancements to SPF to consider constraints: Constraint-Based

Routing (CSPF): Path Calculation
 Explicit route selection
 Bandwidth parameters and recovery mechanisms defined
 Connection Admission Controls (CAC) enforced
 (policing, marking, metering, scheduling, etc)

• Enhancements to the Signalling Protocols to support explicit

constraint-based routing: Path Creation
 LDP  CR-LDP
 RSVP  RSVP-TE
Slide 11
What’s involved in information
distribution to support TE?
• Information distribution is broken down into three
pieces:
 What information is distributed and how to configure it
 When information is distributed and how to control
when flooding takes place
 How information is distributed (protocol-specific
details)

Slide 12
What information Is Distributed?
• The idea behind MPLS TE is to allow routers to build
paths using information rather than the shortest IP path.
But what information is distributed to allow the routers to
make more intelligent path calculations?
 Examples:
 Path that has enough bandwidth, special attributes, low delay, …
 Generally, information that has to do with TE objectives/requirements

• MPLS TE works by using OSPF or IS-IS to distribute

information about available resources. Three main pieces
of information are distributed for each link/interface:
 Available bandwidth information, broken down by priority to allow
tunnels to preempt others
 Attribute flags
 Administrative weight
Slide 13
Available Bandwidth Information
• A key feature of MPLS TE is the capability to reserve bandwidth
across the network
• How much bandwidth to allocate to the interface?
 Also depends on oversubscription policies and the policy to enforce them
 Cisco default is 75% of the link bandwidth

• Main elements: interface, allocated, max, percentage. Example:

 P04/2 233250K 466500K 50
• Need to keep track of currently allocated bandwidth to obtain currently
available or reservable bandwidth
• Need both the per-interface and the per-tunnel (TE LSP) bandwidth
 Why both?

Slide 14
Tunnel Priority
• Some LSPs or tunnels are more important than others.
For example, tunnels for voice traffic.
• Need capability to allow tunnels to preempt others.
 Each tunnels has a priority
 Lower-priority tunnels are pushed out and are made to recalculate
a path, and the resources are given to the higher-priority tunnel
 8 priority levels (0-7)
 Destructive to other tunnels, use only necessary
 In a real network, the preempted tunnel can have an alternative
path for backup and the tunnel will come up
 Example

Slide 15
Setup and Holding Priority
• Each tunnel actually has two priorities – a Setup priority
and a Hold priority (RFC 3209)
• Idea is to use Setup priority to decide whether to admit the
tunnel, Hold priority is used to compare priority if
competition comes along for a new tunnel
 Usually treated the same, but can be different
 Application: once the tunnel is setup, the Hold priority could be set
to the highest, which means that it cannot be preempted by any
other tunnels.
 But Hold priority must be >= Setup priority, why?

Slide 16
Attribute Flags
• MPLS TE allows you to enable attribute flags.
• An attribute flag is a 32-bit bitmap on a link that
can indicate the existence of up to 32 separate
properties of that link.
 ISPs have the freedom to manage these bits
 Example:
 Assuming 8-bit and a link that has attribute flags of 0x1 (0000
0001) means that the link is a satellite link.
 If you want to build a tunnel that does not cross a satellite link,
you need to make sure that any link the tunnel crosses has
the satellite link bit set to 0
 Need a mask

Slide 17
Administrative Weight or Metric

• For MPLS TE, two costs are associated with a

link – the TE cost and the IGP cost.
 Allow to present the TE path calculation with a different
set of link costs than the regular IGP SPF sees.
 In other words, you can change the cost advertised for
the link, but only for traffic engineering. Why?
 Useful in path calculation. Examples:
 Networks that have both IP and MPLS TE traffic
 Delay-sensitive link. Example: OC-3 land line and OC-3
satellite link have different delays, but with the same
bandwidth.

Slide 18
When Information Is Distributed?

• IGP floods information about a link in three cases:

 When a link goes up or down
 When a link’s configuration is changed (e.g., link cost)
 When it’s time to periodically flood the IGP information

• For MPLS TE, there is more to consider:

 When link bandwidth changes significantly
 Link attribute(s) changed

• What is “significant”?

Slide 19
What is Significant?
• How to define significant?
 Percentage of link bandwidth
 Is it enough?
 Rules are different for every network, situation, and link
 Cisco uses default flooding thresholds (15, 30, 45, 60, 75, 80, 85,
90, 95, 96, 97, 98, 99, 100) on links. If the thresholds are crossed,
link bandwidth is flooded.
• Flood insignificant changes periodically
 If available bandwidth has changed and it hasn’t been flooded,
the changes will be flooded every 3 minutes (default value, but
configurable), more frequently than IGP refresh interval
• If error, flood immediately
 A path setup fails due to lack of bandwidth. Available bandwidth
has been changed since the last time flooding occurred.
• Should be considered in TE methods.
Slide 20
How Information Is Distributed?

• MPLS TE in OSPF, hence OSPF-TE

• MPLS TE in IS-IS, hence ISIS- TE
• MPLS TE enhancements and IP-Extended TLVs are
closely related.
 Type 1: router address TLV: MPLS TE router ID
 Type 2: link TLV: 9 sub-TLVs
 Link type, link ID, local I/F IP addr, remote I/F IP addr, TE metric
(cost, admin-weight), max link bw, max reservable bw,
unreserved bw (per priority), attribute flags.

• Before you can do MPLS TE, support for wide metrics

must be enabled.

Slide 21
Constraint-Based Routing

Slide 22
Constraint-Based Routing
• Parameters over and above “best effort” are
constraints
 Constraint = order in which LSRs are reached
 Constraint = description of traffic flow, bandwidth,
delay, class, priority
 Constraint = edge traffic conditioning functions such as
marking, metering, policing, and shaping
 Constraint = Recovery mechanism for “protection” of a
working LSP
• Supports and enables QoS/CoS functions for;
 IP DiffServ and IntServ
Slide 23
Constraint Route Signaling
Operational Model
OSPF-TE
IS-IS-TE
Operations Performed by the
Ingress LSR
Traffic engineering
Routing table
Database (TED)

1) Store information from IGP flooding

2) Store traffic engineering information

OSPF and IS-IS - TE Extensions

Distributed (piggybacked) on Opaque Link State Advertisements
Encoded as new Type Length Values (TLVs)
Metrics: Bandwidth, Unreserved Bandwidth, Available Bandwidth, Delay,
Delay-Jitter, Loss Probability, Administrative Weight, Economic Cost

Slide 24
Constraint Route Signaling
Operational Model
OSPF-TE
IS-IS-TE
Operations Performed by the
Ingress LSR
Traffic engineering Constrained User
Routing table
Database (TED) Shortest Path First Constraints

1) Store information from IGP flooding

2) Store traffic engineering information
Explicit route
3) Examine user defined constraints
4) Calculate the physical path for the LSP - CSPF
5) Represent path as an explicit route
Signaling
6) Pass explicit routing to RSVP-TE or
CR-LDP for signaling

Slide 25
Constraint Route Signaling
Egress
LSR

Ingress
LSR

User defined LSP

constraints

• Operator configures LSP constraints at ingress LSR

 Bandwidth reservation
 Include or exclude a specific link(s)
 Include specific node traversal(s)
• Network actively participates in selecting an LSP path that
meets the constraints

Slide 26
Constraint Route Signaling
Example

Seattle

Chicago New
San York
Francisco
Kansas
City
Los Atlanta
Angeles

label-switched-path SF_to_NY { Dallas

to New_York;
from San_Francisco;
admin-group {exclude green}
cspf}

Slide 27
Constraint Route Signaling
Example
label-switched-path madrid_to_stockholm{
to Stockholm;
from Madrid;
admin-group {include red, green}
cspf}

Stockholm
London

Paris Munich

Madrid Geneva

Rome

Slide 28
Signaling Mechanisms
• LDP Label Distribution Protocol
• CR-LDP Constraint-Based Routing - Label
Distribution Protocol
• RSVP-TE Extensions to RSVP for Traffic
Engineering
• BGP-4 Carrying Label Information in
BGP- 4

Slide 29
Constraint-Based Routing
• CBR could be very challenging and complicated.
 Example: need to deliver 60 bricks with only one bike.
 Solution?

 If > 1 constraint:
NP-complete

Slide 30