[c-nsp] 6500/7600 TCAM Usage

[Gert Doering]

Hi,

On Fri, Jun 03, 2016 at 08:50:50AM +0200, Patrick M. Hausen wrote:
> Now that we are planning to replace our supervisor engines
> (3BXL) with VSS capable ones (10G-3CXL) I'm pondering
> to repartition TCAM for 768k IPv4 and 128k IPv6 and
> to go back to full tables.
> Of course monitoring the usage closely. ;-)

I'd stick to "only partial table" - while the XL TCAM is big enough for
"more", the CPU is still slow, and full BGP on that box is stretching
the limits quite a bit (we have a few still running with ~450k v4 routes,
and peer restarts do cause too much CPU load for my taste).

> What puzzles me is: how do vendors go about that in
> the long run? I have been using my search engine of
> least distrust to no avail. Which platforms offer vastly
> bigger TCAMs, like at least twofold, better an order
> of magnitude?

ASR9k goes to 4M prefix... plus incredibly fast BGP implementation.

It has other warts, of course, like "it's a router, so it has few ports
and those are expensive".

[..]
> Or can one get around those rather arbitrary hard limits
> completely? Is it possible to e.g. have a TCAM with timestamps
> associated to entries, so one can employ a TCAM as
> a route cache in LRU fashion and process-switch everything
> new/unknown?

It would certainly be possible.  Would vendors be interested in spending
money to let you run their old and now cheap coming from the second-hand 
market gear longer?  Answer yourself :-)

gert
-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 291 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20160603/37c7daf1/attachment.sig>