From: Vaughan Lee (vaughan.lee@xxxxxxxxxxx)
Date: Mon Dec 13 1999 - 07:58:22 GMT-3
The thing you need to note in this example that it is using route-reflectors
with peer-groups.
Part of a peer-group's function is to reduce the number of messages that
need to be computed by the router i.e. one message is formed and sent to all
peer-group members. If I remember rightly, the main issue that arises from
this is that when a route is passed from one bgp peer to another, the route
is poisoned in reverse via a WITHDRAWN. In the example, routes from RTE to
RTF would be poisoned by RTF, but as one WITHDRAWN message is made and sent
to all peers, the poison would also get passed to RTD, so causing a routing
problem.
The way around this is to set up bgp peers between RTE and RTD as well as
RTE / RTF and RTD / RTF, and then set the 'no bgp client-to-client
reflection' on RTF.
Hope this is a bit clearer - I've tried to keep it brief. It's explained
more in the sub-chapter 'Route Reflectors and Peer Groups' on pages 261 -
263.
Vaughan Lee, Network Specialist.
CCIE #5250
debis IT Services (UK) Ltd.
E-mail: vaughan.lee@debis.co.uk
Direct Phone: 01908 279561
Direct Fax: 01908 279061
http://www.debis.co.uk/
-----Original Message-----
From: sono@softbank.co.jp [mailto:sono@softbank.co.jp]
Sent: 13 December 1999 10:22
To: kongck@rocketmail.com; ccielab@groupstudy.com
Subject: RE: Route Reflectors in Halabi's book
kongck,
Then what's good about using route reflectors
if a client has to establish peers 'both' with the
RR and the other client, in case of a cluster
consisting of three routers including one RR.
Isn't it a full-mesh ?
I tried this in my lab after my first post.
It looks working well without having a peer
between the two clients.
sono
-----Original Message-----
From: kongck [mailto:kongck@rocketmail.com]
Sent: Monday, December 13, 1999 8:03 AM
To: sono@softbank.co.jp; ccielab@groupstudy.com
Subject: Re: Route Reflectors in Halabi's book
no,both peer will be established,
--- sono@softbank.co.jp wrote:
>
> Hi guys,
> One question about BGP route reflectors.
> In Halabi's book on page 424,
> RTD, which is a client of a route reflector RTF,
> has a peer with RTE, which is the other client of
> the same route reflector RTF by the following
> configs:
>
> neighbor 172.16.25.2 remote-as 3
> neighbor 172.16.25.2 next-hop-self
>
>
> In my understanding, a client needs to establish
> a peer only with its route reflector. (Otherwise it
> doesn't
> make sense to use route reflectors.)
> What am I missing?
>
> Regard.
>
> SONO
>
>
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:21:59 GMT-3