From: Peter van Oene (pvo@xxxxxxxxxxxx)
Date: Thu Mar 21 2002 - 14:12:36 GMT-3
Quick minor correction. I actually misread the topology and thought it was
A reflects to B who reflects to C. However, I stand by my point that this
isn't a bad thing. The only caveat is that B, or R2 in your case, will
need full routing information. If you are filtering the reflected routes
in anyway (ie reflecting only default for example) R2 may drop traffic
destined to R3. In most transit AS's, we're dealing with routers will full
views and this wouldn't be an issue.
At 10:38 AM 3/21/2002 -0600, Diment, Andrew wrote:
>This would not be a good design. One of the purposes for route reflectors
>is so you don't have to make a fully meshed bgp peered network when the
>physical network is not fully meshed. Like in a hub and spoke setup, the
>hub should be the reflector not one of the spokes. If the spoke was the
>reflector and it went down, all of BGP is toast.
>
>-----Original Message-----
>From: Joe Higgins [mailto:netsat@optonline.net]
>Sent: Thursday, March 21, 2002 10:07 AM
>To: ccielab@groupstudy.com
>Subject: bgp route reflection
>
>
>Is the following bgp route reflector configuration valid and/or good
>practice?
>r1, route reflector, > r2, route reflector client, > r3 route reflector
>client. I have read that the route reflector topology should match the
>physical topology of the network. In the above example it does not
>match as router r3 must go through r2 to get to its route reflector,
>r1. Any insight will be appreciated..
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:57:16 GMT-3