In most provider
networks, BGP route reflectors will be used to ensure that advertised NLRIs reach all
PEs in the network.
Each PE learns all BGP NLRIs with VPLS IDs matching locally configured VPLS
instances and with RTs matching configured RTs for those VPLS instances. For each
learned NLRI, the PE then uses FEC 129 signalling to establish a pseudowire to the
BGP next-hop for the NLRI (i.e., the loopback address of the PE that advertised the
NLRI). The AGI is set to the VPLS ID, the SAII to the PE identifier portion of the local
NLRI (i.e., the NLRI with the RD stripped off), and the TAII to the PE identifier of the
learned NLRI. This ensures that exactly one pseudowire will be established between
any two VSIs with matching VPLS identifiers and RTs.
Configuring separate values for the VPLS ID, the import/export RTs (i.e., the RTs to
advertise and the RTs to learn NLRIs from) and the RD are, in most cases, unnecessarily
complex. Implementations may choose to simplify configuration by automatically
creating RDs and RTs based on the configured VPLS ID.
VPLS Scaling The goal for VPLS is to provide a scalable Ethernet service. However, it
is important to remember that there are intrinsic limits to scaling a single Ethernet
LAN, and since VPLS emulates an Ethernet LAN, it is impossible to overcome these.
Pages:
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009