From: Peter Van Oene (pvo@xxxxxxxxxxxx)
Date: Tue Jan 09 2001 - 15:47:35 GMT-3
Your very correct. Keep in mind that most people don't want fully meshed IBGP
due to the extreme amount work required to operate and scale such a network. R
R's offer a means of scaling to large size BGP domains.
Pete
*********** REPLY SEPARATOR ***********
On 1/9/2001 at 11:00 AM Ronnie Royston wrote:
>I understand one of the main purposes of using route reflectors to be for
>providing iBGP peers a means to share iBGP learned routes. In my
>understanding, BGP is built so that iBGP peers will not forward iBGP learned
>routes, only eBGP learned routes. iBGP peering should be fully meshed so
>routing information is not lost. In case you do not have a fully meshed
>topology, route reflectors provide a means to override the default behavior
>of BGP, allowing iBGP servers to forward iBGP learned routes to clients.
>
>The senario you described:
>
> R7--R1 R4
> \ / \ / \
> R2 R3 R5 R6
>
>is a bit confusing to me. The point of route-reflectors is to allow the
>sharing of iBGP learned routes with other iBGP neighbors without having a
>fully meshed network. Redundancy is an option, but be sure you actually
>have the physical redundancy to support your logical redundancy. I hope
>that helps. Good luck.
>
>
>-----Original Message-----
>From: Lykourgiotis Paraskevas [mailto:ParaskevasL@pcsystems.gr]
>Sent: Tuesday, January 09, 2001 8:26 AM
>To: ccielab@groupstudy.com
>Subject: Bgp Route Reflectors ('tree structure...'?)
>
>
>Hi all.
>
>Do you know if it is possible in BGP to have this configuration :
>
>R1 has : R2 and R3 as RR clients
>R4 has : R5 and R6 as RR clients
>R7 has : R1 and R2 as RR clients
>
>Thanks
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:27:25 GMT-3