From: Joseph Brunner (joe@affirmedsystems.com)
Date: Thu May 22 2008 - 04:44:12 ART
The really strange thing is it's selecting the IBGP route over EBGP which is
what I want but from my understanding this shouldn't be happening. I've seen
this happen both on [ CE1 ] and on [ PE1 ]. How is this possible?
>BGP selects the BEST BGP path from among Internal and External paths. BGP
>will run its selection algorithm, and install the winner in the routing
>table using the AD of the winning protocol- IBGP or EBGP
I found some info which said that in this scenario the normal BGP bestpath
process would be used instead of comparing the AD first. This is the first
I've heard of that and would like ot know if it's true?
>and you have a lab date schedule, right? Tell me you have life insurance
>when you drag race drunk?
I mean what's the point of the backdoor command or messing around with
distance if that's the case in this scenario.
>the "backdoor command" is to prefer an IGP with a lower AD over EBGP.
>EBGP's AD will be raised to 200 so OSPF/EIGRP will be it...
>the "backdoor command" is also good for when your neighbors like to see
>young, attractive drunk women stop in and our of your house in the stix
I tried changing local pref and the as-path and it seems to be used over AD.
Has anyone else seen this before?
>again AD is a function of the final job of BGP- selecting the best path for
>installation to the routing table and to advertise that path to neighbors.
>the winning path goes in the routing table with AD 200 if learned from
>an IBGP peer, and 20 if it was an EBGP peer. I mean after all, how do think
>local preference, MED, and any other AS wide attribute work? They do
>need to be "EBGP" routes, which with your old logic, would always win on
>AD. For absolute local control, we have WEIGHT ;)
-Joe
Light saber builder #19366
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Michael Whittle
Sent: Thursday, May 22, 2008 2:17 AM
To: CCIE Lab Group Study
Subject: EBGP vs. IBGP decision process "problem"?
Hi all,
I saw something strange the other day and I'm not sure if it's a problem or
expected behavior. My understanding is after a longest match is selected
from the routing table if there are two equal routes from two different
protocols it will select the one with the more preferred AD. Well I saw
something the other day which contracted this with EBGP and IBGP. I've seen
it now in two completely different scenarios using different kit, software
and technologies.
In my lab I have the following setup:
___mpbgp____[ MPLS VPN ] ___mpbgp___
| |
[ PE1 ] [ PE1 ]
| |
ebgp ebgp
| |
[ CE1 ]---ibgp---[RRCE1]-----[RRCE2]---ibgp---[ CE2 ]
| |
[ LAN1 ] [ LAN2 ]
--- [ LAN1 ] now needs to get to [ LAN2 ] and I want it to use the backend link. By default I would have expected it to route out via [ PE1 ] because it would be receiving the route via the EBGP via the MPLS VPN with an AD of 20 and it would be receiving the same route via IBGP from [RRCE1] with an AD of 200. The default behavior without a backdoor link or manually changing the distance should route the traffic out via EBGP (20) instead of IBGP (200). The really strange thing is it's selecting the IBGP route over EBGP which is what I want but from my understanding this shouldn't be happening. I've seen this happen both on [ CE1 ] and on [ PE1 ]. How is this possible?I found some info which said that in this scenario the normal BGP bestpath process would be used instead of comparing the AD first. This is the first I've heard of that and would like ot know if it's true? I mean what's the point of the backdoor command or messing around with distance if that's the case in this scenario. I tried changing local pref and the as-path and it seems to be used over AD. Has anyone else seen this before?
Regards, Mike
This archive was generated by hypermail 2.1.4 : Mon Jun 02 2008 - 06:59:18 ART