However, the disadvantage of using the existing
format is that in order for the PE to know from which pseudowire a packet has been
received, it must assign a separate PW label to each peer PE (since there is no information
in the pseudowire payload to identify the sender and the tunnel label may be
stripped by the penultimate hop or may not uniquely identify the sender, for example,
when multipoint-to-point LDP is used to establish the tunnels). Typically, raw mode
Ethernet pseudowires are used for VPLS rather than tagged mode pseudowires. This
is because the devices that implement VPLS generally have full Ethernet switching
functionality, including the ability to add and remove VLAN tags.
When a frame arrives on a VPLS attachment circuit, the PE learns the source address
of the incoming Ethernet frame and, if needed, adds it to its VPLS FIB. The PE
then looks up the destination MAC addresses in the FIB. If the destination MAC is
known in the FIB, then the frame is forwarded to the indicated attachment circuit or
pseudowire. If the destination MAC is unknown (or is broadcast or multicast), then the
PE replicates the packet and sends one copy to each other local attachment circuit and
to each pseudowire.
Pages:
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004