TABLE OF CONTENTS Depending on the object of protection TE link or service path one can identify two different levels of recovery.
Link level recovery (also referred to as span recovery)
Path level recovery
In the context of this book we define a span as an abstraction associated with network resources that are necessary to deliver user traffic between a pair of nodes that are adjacent in the data plane. From the traffic engineering perspective, a span can be represented (and advertised) as a separate TE link. Alternatively, several spans interconnecting the same pair of nodes can be collectively represented as a single (bundled) TE link. See more details on TE link advertising in Chapter 8.
In the case of span recovery, all services traversing a particular span are protected against any failure detected on the span. Specifically, when the span fails, all span-protected services are switched simultaneously onto some other parallel span (that is, a span interconnecting the same pair of nodes as the failed one). It is worth emphasizing that not all protected services are recovered through this operation only those that are provisioned to be span-protected. In the span recovery model the deciding entities are located on the span's ends, and the recovery operation is triggered either by local failure detection or by a failure notification message received from the opposite end of the span. We will discuss span recovery in Section 7.6.
In the path level recovery model each
