[c-nsp] Routing design question

[Gert Doering]

Hi,

On Wed, Sep 26, 2007 at 11:55:03PM -0700, Bruce Pinsky wrote:
> >>We have a handful of static routes (less than 40) that range from /25 to 
> >>/30's connected to each 6503.
> >
> >That's what I'd do.  Add UDLD to detect a one-way fiber connection.
> >
> >The routing protocol will also make sure that you really have end-to-end 
> >connectivity, and not just a link.
> 
> Well, if it's truly gigE, then auto-negotiation will be more responsive 
> to failures than UDLD.  

Will auto-neg signal one-way fiber failures (after the link has already
been brought up and autoneg'ed successfully)?  Never tried that.

> If supported in his version, BFD would also be 
> more responsive than UDLD.

I'm not sure whether there is BFD on Sup2/MSFC2, but if there is, and 
it's reliable these days, indeed.

> Relying on routing protocol timers is a poor substitute for proper link 
> failure detection.

As far as I understood the question, it wasn't meant so much as "how to 
get millisecond failover" but "how to set this up so that it's easy to
understand, diagnose, and reliable in the face of typical failures".

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