TABLE OF CONTENTS One can distinguish the following stages of a service recovery process.
Fault detection
Fault hold-off
Fault localization
Fault correlation
Fault notification
Recovery operation
Traffic recovery
The five first stages are collectively called fault management. Fault detection is the only stage that cannot be accomplished without interaction with the data plane all other stages can be realized by either the control plane or the data plane independently or by cooperation between the two planes.
When a fault occurs, the nodes adjacent to the failed link or node do not detect the fault instantly. It takes some time for the hardware dedicated to fault detection (for example, components monitoring signal overhead bytes, power level, and so forth) to identify the failed component, define the exact fault state, and notify the entity responsible for fault management (for example, the GMPLS control plane). The fault detection time may depend heavily on the data plane technology (for example, TDM or WDM), how intrusive each node is in the processing of signals (for example, whether electronic components are used for timing synchronization, signal regeneration, or signal monitoring), and whether the signal is normally expected to be continuous or only present when data is being transferred.
The fault management entity does not usually react to the fault notification right away. First, it moves the service into the fault hold-off state. This is necessary because a lower network layer may have a recovery scheme of its own, and it is...
