This week's tip will focus on MPLS traffic engineering (TE). MPLS TE allows organizations to design robust, resilient, optimized backbones for delivering mission critical application services.
In previous tech tips we have discussed the creation of LSPs over the MPLS backbone. Remember that an LSP is just a logical circuit created via the layer 3 path between source and destination endpoints on the MPLS cloud. The logical circuit path follows the routed path determined by the Interior Gateway Protocol (IGP). The requirements for diversity and rapid convergence around failures remain a critical issue in MPLS environments as well. This requires multiple paths between end destinations and mechanisms for load sharing and rapid failover. In the past the load sharing was handled by manipulating the interior routing metrics and the rapid failover was provided by layer 2 technology (eg Sonet 50ms failover). In the MPLS world, everything is done at layer 3 and as such there needs to be a mechanism for providing these capabilities.
MPLS utilizes traffic engineering mechanisms to control the path and backbone network resources consumed by LSPs. TE principles can be deployed within the backbone to manipulate when, if, and how the LSPs are set up and to control what traffic goes on what LSP. As I mentioned previously the LSPs are set up to follow the routed path and as such IGP route metric manipulation is one way of altering the LSP path architecture. However, this relies solely on the routing protocol in use, which does not provide any information on the characteristics of the traffic nor any network capacity restraints that may impact the delivery of the traffic. MPLS traffic engineering offers the ability to manipulate what traffic traverses which LSPs as well as provide a path that meets the network capacity requirements for a data flow across the backbone.
MPLS can utilize either RSVP or CR-LDP to provide the required TE capabilities as discussed above. For quite some time there were major discussions within the industry and vendor camps as to which of these to use. As of today, the CR-LDP development has ended (for now), and RSVP is emerging as the de-facto standard for implementing MPS TE.
RSVP is essentially a signaling protocol that allows the network to reserve paths based on metrics such as bandwidth required, traffic conditions (e.g. QoS), or a predefined path. If the network can meet the traffic's requirements, the path is set up and the traffic is forwarded. The path is represented by the multiple LSPs that are created to forward the traffic.
RSVP is defined on the MPLS backbone and allows one to clearly define what traffic goes where based on available network resources. This is very similar to how ATM utilized signaling protocols to establish virtual circuits across the backbone. This can now be done at the routing layer. The next tech tip will focus on some of the ways that RSVP can be used to enable traffic engineering.
Robbie Harrell (CCIE#3873) is the National Practice Lead for Advanced Infrastructure Solutions for SBC Communications. He has over 10 years of experience providing strategic, business, and technical consulting services to clients. Robbie resides in Atlanta, and is a graduate of Clemson University. His background includes positions as a Principal Architect at International Network Services, Lucent, Frontway and Callisma.
