Solving the VPLS Ingress Replication Problem
One major issue with VPLS is that it uses ingress replication to flood broadcast, multicast,
and unknown traffic from the ingress PE to each egress PE. Although this issue
is ameliorated somewhat by H-VPLS (in that replication now occurs at the ingress and
466 Chapter 14
egress N-PEs, and no replication at all at the ingress U-PE), it still presents a major
challenge to providers who wish to support IP multicast traffic over VPLS instances.
As noted in ???Triple Play Aggregation,??? there are topology-specific workarounds that
may be used to provide efficient support for IP multicast; however, many providers are
looking for a generic solution to the problem.
The multicast VPLS solution makes use of mLDP and/or P2MP RSVP-TE to build
multipoint LSPs to carry broadcast, multicast, and unknown VPLS traffic (though
unknown unicast packets may optionally be replicated at ingress). For each VPLS, there
may be one ???inclusive tree??? that may be used to send traffic to all PEs in the VPLS (in
fact, multiple VPLS instances may share one inclusive tree to reduce multicast state in
the network core).
Pages:
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051