[c-nsp] Compression on MPLS

[Jared Mauch]

On Sat, Dec 04, 2004 at 10:37:25PM +0100, Kinczli Zoltán wrote:
> hello,
> 
>  just a theoretical Q:
> 
>  couldn't compression be done in CEF process-path?
> CEF: so the labels could be determined from the CEF table
> process path: still punted, so that the CPU can do it's compression task?
> 
>   on 12.2M being old: it's the youngest GD code ;)

	At least he didn't tell you to load 12.3 on a GSR or other
sillyness.

	The way cisco calls code 'old' is quite weird, because they
rarely take into account the train, platform and other important
things that are going on..

> -----Original Message-----
> From: Oliver Boehmer (oboehmer) [mailto:oboehmer at cisco.com]
> Sent: Saturday, December 04, 2004 1:08 PM
> To: kerwin m; signal at shreve.net; cisco-nsp at puck.nether.net
> Subject: RE: [c-nsp] Compression on MPLS
> 
> 
> 
> > I have a FR backbone on which we are running MPLS so you can say its
> > MPLS on FR , we are trying to run compression (stac) on the points
> > between the CPE & the PE but no sooner do we run STAC the routes
> > dissappear from the VRF table ,any idea on how to run compression in
> > MPLS
> 
> unless something chanaged quite recently (but you're running 12.2M, so
> rather old code), stac compression will punt the packet to the process
> path, which usually breaks cef/mpls-vpn.

	This sounds like a bug waiting to be filed :)

	there's no reason that process switched packets
shouldn't work, except on fully distributed platforms (eg: gsr) where
there is no software path.  At minimum the software path should always
work.

	- jared

-- 
Jared Mauch  | pgp key available via finger from jared at puck.nether.net
clue++;      | http://puck.nether.net/~jared/  My statements are only mine.