Re: BGP versus OSPF/ISIS

From: Ahmed Mustafa (ahmed.mustafa@sbcglobal.net)
Date: Sat Feb 28 2004 - 16:51:40 GMT-3


Thanks Scott !

I know there is no trick, but why route would only appear under ospf
database in r3 when the ebgp link between r2 and r6 is down.

It just clicked and I think I know my answer now. Correct me if I am wrong.

Here is what exactly happening.

If the ebgp link between r2 and r6 is down, then obviously r2 learns the
loopback of r6 via ISIS. r2 is redistributing ISIS into ospf, and then the
route appears in ospf database as external route.

When I bring the ebgp link up between r2 and r6, r2 learns the route from
bgp, and the route disappears from the ospf database since ISIS is out of
the picture.

Regards,

Ahmed

----- Original Message -----
From: "Scott, Tyson C" <tyson.scott@hp.com>
To: "Ahmed Mustafa" <ahmed.mustafa@sbcglobal.net>; <ccielab@groupstudy.com>
Sent: Saturday, February 28, 2004 10:07 AM
Subject: RE: BGP versus OSPF/ISIS

It's an easy answer to what you are looking at. eBGP has the lower
administrative distance so it will always believe that route over the
IGP. So you have a design issue here and how do you want to resolve it?
Decide whether you want the route in IGP or EGP. Even though the
administrative distance is 200 to router3 for R2 it is 20 from R6. So
since this is his best route, to R3 perspective, this is what he is
going to advertise to his neighbors. Remove it from one or the other.
If you choose to advertise it via eBGP then make sure for the neighbor
relationship in the same AS use the next-hop-self for the neighbor
relationship with R2 and R3 so R3 will know how to get to the loopback
of R6 if you are not redistributing the transit link between R2 to R6.
There are no tricks here and it is working exactly as it should be.

HTH

-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Ahmed Mustafa
Sent: Saturday, February 28, 2004 3:22 AM
To: ccielab@groupstudy.com
Subject: BGP versus OSPF/ISIS

Can someone look into this

R6 connected to R2 via ISIS level 1

R2 connected to R3 via HDLC link/ISIS Level 2
R2 connected to R3 via Frame-Relay link/OSPF-----> R2 is also doing
mutual
redistribution

I announced r6 loopback interface 132.1.6.0/24 to ISIS via redistribute
connected.

R3 learned that route obviously through Frame-Relay due to lower admin
distance. I then changed the admin distance of ISIS to 106, and r3
installed
this route via ISIS. So far so good

Now comes BGP, and totally screwed this nice scenerio above

R6 is in AS 100
R2 and R3 are in AS 300

I redistributed r6 loopback now into BGP via redistribute connected,
not
through network command.

r2 will obviously install this route as a BGP route because of lower
adminstrative distance External BGP 20 over ISIS 115.

Now r3 learns that route via BGP as well with the distance 200 since r2
and r3
are in same AS. I thought that the route 132.1.6.0 should have learned
through ISIS or OSPF if I removed the admin distance command of ISIS
from 106
to default 115 because both OSPF and ISIS have lower admin distance than
IBGP
200.

My guesses:

The route learned via BGP since it is in r2 routing table regardless of
r2
should have this route in ISIS database as well as in OSPF database
since r2
is also doing redistribution.

Issues: If I shut down the BGP link between r2 and r6, bingo, the route
will
appear in OSPF database as an external route on r2, but when the BGP
link is
up between
r2 and r6, the route will disappear from OSPF database.

Any ideas !!!

Sorry for the tedious email, but Group can learn from this and also
advise.

Regards,

Ahmed



This archive was generated by hypermail 2.1.4 : Fri Mar 05 2004 - 07:13:59 GMT-3