Advanced Protection and Restoration: Protecting the Service
Chapter 3, Protection and Restoration in MPLS Networks, described the mechanisms available for providing 50 ms recovery for link and node failures on an LSP. In Chapter 6, we saw how these mechanisms can be extended for P2MP LSPs. In this chapter, we look at various services offered over MPLS networks and see how similar recovery times can be achieved at the service level. We pay particular attention to failures at the egress points of an LSP, because they are not covered by the FRR mechanisms described so far. We also look at various methods for protecting the service, and describe a novel approach for LSP tail-end protection.
Familiarity with MPLS FRR (Chapter 3), LSP nesting and stitching (Chapter 5) and VPNs (Chapters 7 and 8) is assumed for understanding this chapter.
14.2 THE BUSINESS DRIVERS
The first question to ask is why would 50 ms recovery times be required for MPLS services. To answer this, remember that the requirement for 50 ms failure recovery [GR253-CORE] has its origins in the technologies used for transport.1 The MPLS applications that we discussed in the previous chapters, such as pseudowires and VPNs, give the customer the equivalent of a physical transport layer, using a virtual infrastructure. Because the applications that run in the customer's context may have very stringent requirements in terms of loss, it is natural that the customer expects the same 50 ms failure recovery ...