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 998 | Next

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

"Delivering Carrier Ethernet: Extending Ethernet Beyond the LAN"


For example, as hosts are added to an Ethernet LAN, the broadcast and multicast traffic
load typically increases due to hosts issuing periodic ARP requests or searching for
local network resources (this is protocol dependent, of course).
The LANE and MPOA architectures in the 1990s attempted to address Ethernet
scaling, but used a very complex design including dedicated servers for forwarding
broadcast, multicast, and unknown packets. This experiment was not a great success;
hence VPLS adopted a simplified architecture where the PE devices themselves replicate
those packets.
One specific issue when connecting Ethernet switches to a VPLS instance is that the
VPLS learns all MAC addresses for the switched network. This constrains VPLS scaling
as the PE devices have to learn all active MAC addresses in the switched network.
For this reason, many service providers offering VPLS service either limit the number
of MAC addresses per attachment circuit, effectively restricting customers to attaching
routers, or charge customers for blocks of MAC addresses.
MPLS 447
H-VPLS The flat topology of VPLS (where every PE has a pseudowire to every other
PE) causes two major scaling challenges:
?–  The ingress PE has to replicate broadcast, multicast, and unknown frames to all
other PEs in the VPLS.


Pages:
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010