From: Reza Toghraee (reza@toghraee.com)
Date: Wed Oct 29 2008 - 03:16:12 ARST
Nit,
Since you have IGP reach ability to R1's interface connected to BB1 and R3's
interface connected to the BB2,, you will have reachability to prefixes
learned from AS254.
BUT, Imagine that you are asked to load-balance traffic on R2 between links
to R1 and R3.
In this case since the learened prefixes from R1,R3 share common attributes
(Weight, LocalPreference, AS-path, Origin, MED), BGP will select the route
with better IGP metric. Imagine R2 is learning R1's interface connected to
BB1 as OSPF external BUT learning R3's connection to BB3 as Inter-Area. So
R2 always prefer link to R3 to reach AS254.
In this case if R2, has an intra-area connection to R1,R3 you can solve it
by using next-hop-self command on R1,R3.
So, R2 will see both IGP metric types are intra-area and you can fix the
metrics to be equal. Then if you enable the bgp load-balancing for internal
neighbors, BGP will install 2 routes for each prefix learned from AS254.
I saw this along a Checkit LAB.
Reza
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Nitro Drops
Sent: Wednesday, October 29, 2008 8:54 AM
To: ccielab@groupstudy.com
Subject: BGP - Next-Hop-Self
Setup as follows
BB1 (AS254) >> EBGP >> R1 (AS100), R2 (AS100), R3 (AS100) >> EBGP >> BB2
(AS54)
AS100 = Full mesh among R1, R2, R3
There is full IGP reachbility from AS100 to AS254 and AS54.
Qns.) Is there still a need to use 'next-hop-self' on both R1 and R3 to
neighbor R2. So that R2 will learn R1 as the next hop address to reach
AS254,
and R3 as the next hop address to reach AS54 prefixes?
My understanding of using 'next-hop-self' is when there is no IGP routes
from
R2 to BB1 and BB2.
Did an ASET lab, solutions showed using 'next-hop-self' even R2 has IGP
routes to BB1 and BB2.
Cheers
Nit
This archive was generated by hypermail 2.1.4 : Sat Nov 01 2008 - 15:35:23 ARST