Re: [nsp] Couple of CEF oddities

From: Alan Hannan (alan@gblx.net)
Date: Thu Oct 28 1999 - 12:55:21 EDT


  It is a generally accepted perception that
  cisco OSPF does not scale past a certain point,
  and that external routes in OSPF exacerbate the
  overhead issue quite significantly.

  I'd highly recommend you transition to IS-IS.

  -alan

Thus spake Sean Butler (sbutler1@tampabay.rr.com)
 on or about Thu, Oct 28, 1999 at 11:13:51AM -0400:
> We've generally had very good success running CEF,
> even though I have heard others say it does not run
> well at all. Recently, we've come across two very strange
> things and I'm just curious if anyone has seen something
> similar... Code level in both cases is 11.1(26.1CC)
>
> First is an SPF/CEF issue. We have a few routers that
> take about 2 seconds to run SPF on external routes,
> which could actually be the heart of the problem. Most
> routers in the same area 0 run in 1 second or so. When
> CEF is enabled on the routers that take twice as long
> as others to run SPF, then the SPF calc goes up to 6-9
> seconds!!! And the router drops packets to itself and
> transit packets.... :-(
>
> The second issue is one with WCCP and CEF. When
> WCCP is turned on and CEF is enabled, IP INPUT
> goes quite high (as of course doe overall CPU)... Turning
> CEF off and leaving WCCP on drops IP INPUT to
> where we expected it... :-( I know that WCCP can't
> be CEF switched, but I would not expect it to be
> process switched just because CEF is on, instead
> of FAST or OPT switched...
>
> So, anyone see anything else similar? We are of course
> working this through our Cisco team, but I just wanted
> to see if anyone else has had such issues too..
>
> Thanks,
> Sean



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:06 EDT