Thanks to everyone for the input. I've been meaning to read the IOS
Architecture book for some time, and I guess this is yet another reason to
get to it. I guess I'd been a bit too optimistic in my perception of how
distributed "distributed" actually meant, and the role of the RSP for
*IP-*IP forwarding makes a whole hell of alot more sense now.
> The only true distributed switching occurs within the same VIP.
Hrm. In the case of a GEIP/GEIP+, provided you have .1q CEF support, would
this also apply? I guess I need to read the book, but I wonder if there's
a chance that given optimal distribution on a .1q'ed port-channel that you
could optimize to switch each IP independently across vlans. Hrm.
> I have GEIPs running at 400+ Mbps on two different buses with Cybus
> utilization in the 40-50% range with very little to no receive side
> buffering.
That's GEIP and not GEIP+? What's your packet size distribution to get
that kind of throughput?
Thanks again.
..kg..
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:41 EDT