13.3. DETECTING AND TROUBLESHOOTING FAILURES
Detecting the failure is the first step towards fixing it. As MPLS networks start carrying more and more services with strict SLAs, fast failure detection becomes a must. However, given the MPLS fast-reroute mechanisms discussed in the Protection and Restoration chapter (Chapter 3), why are we even discussing failures, let alone their fast detection, in this chapter? The answer is because the local protection mechanisms of MPLS protect against a physical link or a node failure, but other events, such as corruption of a forwarding table entry or a configuration error, can also cause traffic forwarding problems.
When talking about forwarding failures, there are two goals. The first, and most important, is to detect the problem quickly. For a provider, the worst possible scenario is to find out about the existence of a problem from the customer asking why the service is down. The second goal is to automatically recover from the failure. This may mean switching the traffic to a different LSP or even bringing down a service, with the correct indication, instead of blackholing traffic for a service that is reported to be up and running.
13.3.1. Reporting and handling nonsilent failures
From the point of view of failure detection, there are two types of forwarding errors: silent and nonsilent. We will discuss nonsilent failures first, and talk about both detection and fast recovery. Nonsilent failures are the ones that the control plane is ...