Transport network resources do fail. Fibers get cut, and cross-connects, amplifiers, DWDM devices, network controllers, and control channels go out of service unexpectedly. Considering the huge amounts of data carried over transport networks, a single such failure, even for a short period of time, can cause a lot of damage to users of services that happen to traverse the point of failure. Users do not generally tolerate losses of data or connectivity. Usually there is a very stringent requirement on the time within which a service must be recovered after its interruption. The only way for transport Service Providers to meet this requirement is to over-provision their networks so that, at any point in time, active services could be diverted from any potential point of failure onto some other network resources. But even this is not sufficient. There should also be an intelligence that can rapidly detect and localize failures as well as switch the services away from them. Likewise, there should be an intelligence capable of computing diverse primary and recovery paths, so that a single failure will not affect them both. Finally, there should be an intelligence that can place primary and recovery paths of multiple services in such a way that the same network resources could protect multiple services and could be used for so-called extra-traffic services while there are no failures in the network, thereby considerably lessening the cost of over- provisioning without providing a significantly reduced ability to protect against failures.

In this chapter we...