From: Godswill Oletu (oletu@inbox.lv)
Date: Sun May 21 2006 - 11:55:06 ART
Koen,
There are many methods out there that, you can use to achive your stated
objective. One of them that come to mind is to use the 'distance <ad>
<source address>' command on R1 to increase the AD of all routes that R3 is
sending to R1 or use it to reduce the AD of all routes that R2 is sending to
R1; which ever way you choose, R2 routes will make it into the routing table
of R1 and R3 routes will be floating around for redundancy.
Another method is to use the distribute-list on R1 and yank out all R3
routes; well you will also method method your stated objective with this
method, but will loose redundancy.
You can also do some policy based routing on R1, this will create a routing
exception on R1 and allow it to forward traffic according to the policy that
you have setup.
Choose, the method that is most suitable and go for it and there might be
other methods that I cannot think of now, these 3 or 4 methods are a good
place to start off.
HTH
Godswill Oletu
----- Original Message -----
From: "Koen Zeilstra" <koen@koenzeilstra.com>
To: <ccielab@groupstudy.com>
Sent: Sunday, May 21, 2006 8:15 AM
Subject: prefer link in OSPF using multipoint FR interface
> Hi group,
>
> Suppose you have 3 routers.
>
> R1 is the frame-relay hub with a multipoint sub interface. R2 and R3 are
> spokes with a physical FR interface.
>
> R2 and R3 have a shared ethernet segment.
>
> R1-------(FR)-------|
> | |
> | |
> (FR) |
> | |
> | |
> R2----(ethernet)-----R3
>
>
> I want to like to prefer the path for R2 to go via R3 for routes behind
> R1 instead of directly to R1.
>
> I can use ip ospf cost on the physical interface of R2 to make the FR link
> less interesting and prefer the ethernet to R3. However for traffic coming
> from networks behind R1 this won't work. That traffic will still go trough
> the FR link R1-R2 instead of using R1-R3-R2. I cannot set ip ospf cost on
> an interface basis on R1 since that is a multipoint sub. Setting cost on a
> neigbhor is also not allowed for the same reason.
>
> Hope this is clear.
>
> Anybody a solution for this?
>
> thanks,
>
> Koen
>
> -----------------------
> If little else, the brain is an educational toy.
> -- Tom Robbins
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Thu Jun 01 2006 - 06:33:22 ART