From: Godswill Oletu (oletu@inbox.lv)
Date: Sun Dec 04 2005 - 07:32:58 GMT-3
The only reason R3 is preferring the routes from R6 to those from BB3 is
because they are better, the local-preference you setted on R6 made those
routes to be preferred when R3 evaluates them in comparison to the same set
of routes learn from BB3.
If we all agree on that, there are couple of ways to manipulate R3 to
instead prefer BB3 routes without doing any configuration on R6.
Conditional advertisement is one method.
Another method is to apply an inbound filter on R3 that will decrease the
local preference of those prefix coming from R6. You can also use next hop,
set the next hop of the routes coming from R6 to say interface null0 (I have
not labbed this, but if you set the next hop of the routes to an
inaccessible address) bgp will discard the routes and will not add them to
the routing table, this is the first thing the path selection algorithm
does.
You can even play with this further, set the local preference of routes
learn from BB3 to a higher value than the one set on R6.
Chances are depending on your IGP reachability, if you enable
synchronization, those iBGP routes from R6 might disappear.
HTH
Godswill
----- Original Message -----
From: "Josef A" <josefnet@gmail.com>
To: "Group Study" <ccielab@groupstudy.com>
Sent: Wednesday, November 30, 2005 10:29 PM
Subject: A bgp question
> Hello Guys:
>
> Here is a question on bgp next-hop modification.
>
> I have a topology like this:
>
>
> BB3 ------------------ R3
> |
> |
> R5
> |
> |
> BB1------------------- R6
>
>
> BB1 and BB3 are both in AS54.
>
> R3, R5, and R6 are in AS 100.
>
> AS 100 is learning some prefixes from AS54. I have initially set the local
> preference inbound on R6 so that AS100 will prefer the R6-BB1 link to
reach
> those prefixes from AS54.
>
> Thus R3 is now using BB1 as its next-hop for those prefixes.
>
> My goal now is to override that behavior on R3, and force R3 to make BB3
the
> next-hop for those prefixes, without using WEIGHT. I have tried using a
> route-map inbound on R3 matching those prefixes and using the set ip
> next-hop A.B.C.D or the set ip next-hop peer-address command . But this
is
> not working.
>
> What am I missing?
>
> Your thoughts and/or comments is appreciated
>
> Josef.
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Mon Jan 09 2006 - 07:07:50 GMT-3