Cisco Systems 15327 Understanding MPLS-TE

18-13

Ethernet Card Software Feature and Configuration Guide, R7.2

January 2009

Chapter 18 Configuring Ethernet over MPLS

Understanding MPLS-TE

Understanding MPLS-TE

MPLS traffic is normally routed to the least cost path as calculated by OSPF or another IGP routing

protocol. This routing gives little or no consideration to varying bandwidth demands or link loads. MPLS

traffic engineering (MPLS-TE) overcomes this by mapping traffic flows to paths that take bandwidth

demands into account. These paths are known as MPLS-TE tunn els, and they may deviate from the

normal IGP calculated routes.

MPLS-TE (RFC 2702) allow service providers to create traffic engineered tunnels to reserve bandwidth

for specific types of traffic and to provide point-to-point services for end customers. The ML-Series card

supports a maximum of 24 MPLS-TE tunnels. MPLS-TE tunnels can carr y a VC type 5, which tunnels

an Ethernet port, or a VC type 4, which tunnels an 802. 1Q VLAN.

For the ML-Series card to use MPLS-TE, you need to configure three main components. First, you must

implement an IGP routing protocol that conveys and distributes information about the link resources

throughout the MPLS network. For this purpose, the ML-Series card supports OSPF and OSPF-TE

extensions (RFC 2328 and RFC 2370). MPLS-TE extensions for other routing protocols, such as IS-IS,

are not supported on the ML-Series card.

Second, you need to configure a signalling protocol to reserve need ed resources and establish LSPs

across the MPLS network. MPLS-TE tunnels use Resource Reservation Protocol (RSVP) messages

(RFC 2205 and RFC 3209) to accomplish this. The ML-Series card supports RSVP and the RSVP

extensions for LSP tunnels on both POS interfaces and RPR (SPR) interfaces.

For the third component, you need to set up an MPLS-TE tunnel on the appropriate ML-Series card

interface. This requires creating an MPLS tunnel interface with an IP address, destination,

encapsulation, bandwidth, and explicit or dynamic path.

The ML-Series card uses RSVP to establish MPLS-TE tunnels and the associated tunnel labels. Targeted

LDP is still used to establish the VC Labels. Also, RSVP is only used to guarantee the bandwidth on the

intermediate nodes on the tunnel. On the ML-Series card, which will be the end-point of the MPLS-TE

tunnel, RSVP is used only for bandwidth allocation.

You configure bandwidth guarantees on the ML-Series card ports using the Cisco Modular Quality of

Service Command-Line Interface (MQC), just like the standard QoS on the ML-Seri es card. For more

information, see the “EoMPLS Quality of Service” section on page 18-3.

The ML-Series card does not use RSVP messages to carry the information for EoMPLS VCs. LDP

sessions are still used to exchange VC information. Also RSVP does not guarantee bandwidth. It only

allocates bandwidth.

The ML-Series card supports RSVP summary refresh and RSVP refresh reduction (RFC 2961). Refresh

reduction is a set of extensions that reduce the messaging load imposed by RSVP. This helps RSVP scale

to support larger numbers of flows. The global configuration command ip rsvp signalling refresh

reduction enables this feature.

You can configure the ML-Series card to encapsulate and preserve the customer’s Ethernet FCS. The

ML-Series card will carry the Ethernet FCS end-to-end and unmodified across EoMPLS or EoMPLS-TE

tunnels. This end-to-end preservation of the original Ethernet FCS is useful for troubleshooti ng.