SEARCH
0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
Prev | Current Page 999 | Next

Abdul Kasim, Prasanna Adhikari, Nan Chen, and Norman Finn

"Delivering Carrier Ethernet: Extending Ethernet Beyond the LAN"

As the number of PEs in the VPLS grows, the replication
overhead also grows. Because PEs are generally located at the edges of the network,
they may have limited bandwidth available for replicating traffic.
?–  A full mesh of MPLS LSPs, Ethernet pseudowires, and targeted LDP sessions are
required between all the PEs in the VPLS. If RSVP-TE LSPs are used, this may
severely constrain VPLS scaling (see ???MPLS Scaling,??? earlier in the chapter).
The Hierarchical VPLS (or H-VPLS) model addresses these problems by segmenting
the VPLS service into a core of ???hub??? devices that are fully meshed, as in a standard
VPLS, and a number of ???spoke??? devices that are each connected through a single ???hub.???
The hub devices are generally known as Network-facing PEs (N-PEs), and the spoke
devices as User-facing PEs (U-PEs).
In RFC 4762, the N-PEs are referred to as ???PE-rs??? devices, since they must be PE
devices that are capable of routing and switching. The U-PEs are known as ???MTU-s???
devices since they are often located at Multi-Tenant Units and are capable of switching,
but do not need to be capable of routing as they have only a single path into the
network.


Pages:
987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011