From: Omer Ansari (omer@ansari.com)
Date: Wed Sep 04 2002 - 20:47:26 GMT-3
Claudine,
I have faced the same predicament as yours, and this is how i work it.
when faced with IGPs,
first configure all IGPs separately (no redist)
then before redistribution, just sit back and look at the big picture.
even if there is a single point of redist for two protocols e.g.
IGRP<->OSPF on RouterX,
and another point where I'm doing IGRP->EIGRP
say EIGRP is redisting into OSPF somewhere else.
If i dont have a route-map on RouterX for denying IGRP routes coming in
from OSPF, there's nothing to stop RouterX from taking these "ospf
external" routes (which are routes of the IGRP domain)back into IGRP.
so in short, if I am sure OSPF has no chance of learning my IGRP domain
routes from somewhere else, I dont bother with a route-map when redist'ing
OSPf back in.
remember, every routing protocol must save its own butt. intelligent RPs
(OSPF/EIGRP) can assist by providing tagged routes which that routing
protocol can use to deny routes back in.
side note: I've seen eigrp is a better beast, as redist'ed routes have an
AD of 170 and even with no route-maps in place, they dont come back to
bite you in the posterior as the own eigrp routes are AD=90, and are
preferred.
HTH
On Wed, 4 Sep 2002, Claudine DEMAR wrote:
> Hi.
>
> I would like to know whether CCIE candidates are expected to configure
> route-maps to filter redistribution between routing protocols even if we are
> in a one-point-redistribution scenario (there's only one common router
> between the different routing domains).
>
> Should those route-maps be configured systematically even if there's no need
> for them and even if there's no such mention in the lab instructions ??
>
> Thanks,
>
> Claudine
>
> _________________________________________________________________
> Chat with friends online, try MSN Messenger: http://messenger.msn.com
This archive was generated by hypermail 2.1.4 : Mon Oct 07 2002 - 07:43:44 GMT-3