Given the topology attached, is it possible to create a route reflect cluster between the two MPLS routers via different providers? My issue is I need to mesh the BGP between both MPLS routers.
above the config of option C between two MPLS SP, this is my note but it easy to get.
RR is admin by SP
PE is admin by SP
ASBR is admin by SP
so 90% of work done by SP.
This is an iBGP design. I researched more and I believe what I'm looking for is simply creating a client/client peer between the route-servers. I just need to make sure the loopbacks have readability between the two routers through the WAN switch via OSPF. I'll confirm if this resolves the issue.
Hello @thefilmguy ,
you need to configure a standard iBGP session between the two MPLS routers acting as RRS in case the direct iBGP session with a client fails one MPLS router can failover to the session via the other HUB/MPLS router.
Do not configure a cluster-id or you will lose this capability ( or configure a different cluster-id on each by default it is equal to BGP RID of RRS)
Of course this means you need a L3 link between them ( or even two for redundancy) out of interfaces used for the two DMVPN clouds.
a RRS reflects a client advertisement to both other clients and non clients iBGP peers this is why all you need is a standard iBGP session.
Hope to help
Ok i get what you want,
Site a - mpls a <- this ibgp ipv4
Site b - mpls b <- this ibgp ipv4
Between two site we will use ebgp
VPNv4 "since different AS"
This work i already make lab last days with one mpls one as but for you i will do lab for inter AS.
Hello @MHM Cisco World ,
I think the original poster is a customer of two different MPLS ISPs and on each of them he has deployed a DMVPN cloud and its routers speak iBGP over each DMVPN cloud, so it is quite simpler then what you are suggesting that is on the service provider side point of view.
Hope to help
You got it. The idea is to peer the two route-reflectors (DMVPN routers). I'm going to be attempting this shortly. My only question remains is whether or not the cluster-id set globally using the loopback 0 interface is enough or do you have to specifically configure it under the BGP configuration?