When a link or node on the shortest
path for an LSP fails, the upstream nodes will remove the label mapping for the LSP
from their forwarding information base (FIB) until a new shortest path and a matching
label for that path is found. Failure of a link or node may be detected using physical
layer mechanisms (such as loss of light), layer 2 mechanisms (such as PPP keepalives),
or using BFD sessions between adjacent LSRs. The same detection mechanisms may
be used for backup LSPs and for fast reroute (both described below).
Backup LSPs One commonly used protection mechanism is to define a backup path
for an RSVP-TE signaled LSP, ideally disjoint from the primary path signaled for the
LSP. The backup path (often known as a stand-by LSP or as a secondary LSP) may
either be presignaled (so label state is instantiated along all LSR in the backup path)
or simply precomputed (so signalling will be required to re-establish the LSP along the
backup path). When a link or node along the LSP fails, the upstream node will generate
an RSVP-TE PathErr message toward the ingress LSR. The PathErr message needs
to traverse all the nodes and links on the path back to the ingress LSR.
Pages:
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967