5.3. SETTING UP INTERDOMAIN TE LSPs

As discussed earlier in this chapter, the limitation of a TE LSP to a single IGP area is caused by the limited visibility into the topology at the LSP head end. However, once a path is specified, there is no problem signaling it across IGP areas or across ASs. Therefore, the setup of interdomain LSPs is possible without any extensions, e.g. by computing the path offline and specifying the hops at the head end. The problem with this approach is that it forces a service provider to move to an operations model relying on offline computation for both the primary and secondary paths, with the implications discussed in section on offline computation in the TE chapter.[] In addition, the issue of fast reroute is not addressed in this model unless the bypass tunnels protecting the interdomain links are also computed offline.

[] In the case of a multiprovider environment the offline tool would also need to know the TE information of all the links of all the providers involved. That may require, among other things, that one provider discloses its internal topology to another provider, a not very attractive prospect in many cases.

As TE and MPLS-based applications started gaining traction in the industry, more scalable and optimal solutions than the simple setup of an LSP across domain boundaries (whether area or AS) became necessary. The requirements for interdomain traffic engineering [RFC4105, RFC4216], were developed in the TEWG[] in the IETF. The solutions ...

Get MPLS-Enabled Applications: Emerging Developments and New Technologies now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.