RE: [nsp] MSFC2 128,000 route limitation

From: David Sinn (dsinn@microsoft.com)
Date: Tue May 14 2002 - 16:09:48 EDT


"show ip cef sum" will tell you how many routes you have within CEF.
There is no compression between CEF and the FIB. The "compression" come
from the fact that the FIB points to the adjacency table for the actual
next hop information, and thus multiple FIB entries can point to one
adjacency entry.

If you are running Hybrid you can confirm this:

CATOS% sho mls cef
Total L3 packets switched: 124893998234
Total L3 octets switched: 53019378962495
Total route entries: 112579
  IP route entries: 112578
  IPX route entries: 1
  IPM route entries: 0
IP load sharing entries: 295
IPX load sharing entries: 0
Forwarding entries: 112521
Bridge entries: 56
Drop entries: 2

MSFC% sho ip cef sum
IP Distributed CEF with switching (Table Version 86771423), flags=0x0
  112564 routes, 1 reresolve, 0 unresolved (0 old, 0 new)
  112567 leaves, 6888 nodes, 21156688 bytes, 86771426 inserts, 86658859
invalidations
  295 load sharing elements, 96760 bytes, 112359 references
  universal per-destination load sharing algorithm, id 8ADDA64A
  2 CEF resets, 2306608 revisions of existing leaves
  refcounts: 1981829 leaf, 1763584 node

Adjacency Table has 151 adjacencies

David

-----Original Message-----
From: kevin graham [mailto:kgraham@dotnetdotcom.org]
Sent: Tuesday, May 14, 2002 11:59 AM
To: Alex Rubenstein
Cc: cisco-nsp@puck.nether.net
Subject: Re: [nsp] MSFC2 128,000 route limitation

> "... The Catalyst 6500 Family when equipped with a Supervisor Engine 2
can
> provide as many as 128,000 entries in the FIB table ..."

Ran across the same limitation on the 4000/Sup3 and had similar
concerns.
In that case (presumably the same as the 6k), the 128k limitation is
derived from the TCAM, and likewise is only addressable w/ future
hardware
upgrades.

> I am sure this is not news to everyone, but things are starting to get
> close.
>
> We're at 111k routes.

As already mentioned, the limitation is on FIB entries, not routes, but
I
would be curious to know if while doing best-path selection, routes are
aggregated before being entered into the FIB, which would would make the
128k limit much more acceptable still (yeah, still haven't sat down w/
IOS
Architecture)..

..kg..



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:11:57 EDT