>The 10000 remains narrowly focused on the aggregation front in
>terms of line cards, I'm not sure that it's doing anythign that
>couldn't have been done in the GSR framework.
GSR has a lack of packet analyzing features such as ACL, CAR etc. with the current Engine 2 architeture. Engine 3 is coming soon to help in this field on 2.5Gbps/slot cards, Engine 4+/Edge/NextGen will do it someday for 10Gbps/slot cards.
>The other direction is of course the 6500 based "switch" fabric
>with multiple router modules, not all that convincing either.
It shares the central packet processing idea with the 10000, but has a very powerful bunch of ASICs to do the task. Distributed processing can be done on LAN with DFC, it may be extended to WAN in the future.
>Still waiting to see a new generation PA-card based w/distributed
>processing (ala VIP) series materialize.
Unlikely, it would kill GSR... and not needed to compete as the "J-thing" is a centralized architeture.
Rubens Kuhl Jr.
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:44 EDT