The facility bypass case is, of course, much more scalable than one-to-one backup (since
fewer fast reroute LSPs are required), but requires the node that detects the failure (the
Point of Local Repair or PLR) to tunnel traffic for all the affected LSPs by swapping the
outermost label to the one expected by the next hop or next-next hop (the merge point
or MP) and then pushing on a label corresponding to the bypass LSP. This is a good
example of MPLS label stacking in operation. In addition to forwarding traffic over a
detour or bypass LSP, the PLR sends a PathErr message upstream to the ingress LSR,
which may then choose to create a new protected LSP that avoids the failed resource. It
is best practice that the new LSP will be created using ???make before break??? methodology
where traffic is forwarded on the old LSP (and thus over the fast reroute detour or bypass)
until the new LSP has been fully established. Many fast reroute implementations
are capable of protecting 1000s of LSPs within a few milliseconds??”far surpassing the
protection times required in SONET/SDH.
OAM-based Fault Detection The three protection mechanisms mentioned above are
all based on use of the IP/MPLS control plane to propagate fault status.
Pages:
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969