Please don't shoot me, I'm a wireless guy. With the topology diagram attached, I have (2) MPLS service providers that are not sharing BGP. What's the best way to get the MPLS (DMVPN Headend Routers) to share BGP between each other if all routers in the topology are in the same AS?
You didn't share sufficient information to know the purpose and work of BGP here.
You can use iBGP between both MPLS routers. Add a point to point connection or over your LAN as per your availability and design.
We welcome wireless guys to participate in the community and will try to make allowances for limited background. I agree with Deepak that more information would be helpful. Am I correct in understanding that MPLS1 and MPLS2 are different providers? And am I correct in understanding that those providers are not running BGP between themselves? That could be quite a challenge.
If that is the case then perhaps we need to understand what does siteA user MPLS1 for? And what does siteB use MPLS2 for? Can you provide some clarification?
When I see a topology that show a connection between siteA and MPLS1, and a connection between MPLS1 and MPLS2, and a connection between MPLS2 and siteB then I generally assume that BGP between siteA and MPLS1 and BGP between siteB and MPLS2 would support connectivity (and exchange of routes) between siteA and siteB. If that is not the case then perhaps we should consider the possibility of a site to site vpn between siteA and siteB using their connection to the Internet (separate from their connection to the MPLS providers).
So each MPLS is a separate provider. All connecting via iBGP over DMVPN with Cisco IWAN on top. These are the border routers we manage with Cisco IWAN. Because MPLS-CL border router is not advertising learned neighbors from the MPLS-SP border router and vice versa, INET ends up as the best path between the sites. Despite using IWAN, because that BGP route is missing via the MPLS circuit, it has no choice but to go out the INET path.
There is a WAN distro switch behind the border routers, where there routes come in via OSPF. But that doesn't solve the issue where I need Site A to learn the route to Site B via the MPLS connection. So you are correct, we do not have iBGP between the border routers. That's why I am asking if that's possible?
The issue comes down to IWAN is not using the MPLS path to send voice traffic from Site A to Site B because it has no route for that via MPLS, only INET. I classify EF traffic to be sent out MPLS. Does that make sense?
CenturyLink MPLS Border Router
router bgp 65101 bgp router-id 192.168.230.251 bgp log-neighbor-changes bgp listen range 192.168.212.0/23 peer-group MPLS-CL-SPOKES neighbor MPLS-CL-SPOKES peer-group neighbor MPLS-CL-SPOKES remote-as 65101 neighbor MPLS-CL-SPOKES description MPLS-CL Spoke Route Reflector neighbor MPLS-CL-SPOKES update-source Tunnel12 neighbor MPLS-CL-SPOKES timers 20 60 ! address-family ipv4 bgp redistribute-internal network 0.0.0.0 network 10.0.0.0 network 172.16.0.0 mask 255.240.0.0 network 192.168.230.0 network 192.168.230.253 mask 255.255.255.255 network 192.168.231.0 aggregate-address 192.168.0.0 255.255.0.0 summary-only neighbor MPLS-CL-SPOKES activate neighbor MPLS-CL-SPOKES send-community neighbor MPLS-CL-SPOKES route-reflector-client neighbor MPLS-CL-SPOKES next-hop-self all neighbor MPLS-CL-SPOKES weight 50000 neighbor MPLS-CL-SPOKES soft-reconfiguration inbound neighbor MPLS-CL-SPOKES route-map MPLS-CL-IN in neighbor MPLS-CL-SPOKES route-map MPLS-CL-OUT out distance bgp 201 19 200 exit-address-family
Sprint MPLS Border Router
router bgp 65101 bgp router-id 192.168.230.252 bgp log-neighbor-changes bgp listen range 192.168.210.0/23 peer-group MPLS-SP-SPOKES neighbor MPLS-SP-SPOKES peer-group neighbor MPLS-SP-SPOKES remote-as 65101 neighbor MPLS-SP-SPOKES description MPLS-SP Spoke Route Reflector neighbor MPLS-SP-SPOKES update-source Tunnel10 neighbor MPLS-SP-SPOKES timers 20 60 ! address-family ipv4 bgp redistribute-internal network 0.0.0.0 network 10.0.0.0 network 172.16.0.0 mask 255.240.0.0 network 192.168.230.0 network 192.168.230.253 mask 255.255.255.255 network 192.168.231.0 aggregate-address 192.168.0.0 255.255.0.0 summary-only neighbor MPLS-SP-SPOKES activate neighbor MPLS-SP-SPOKES send-community neighbor MPLS-SP-SPOKES route-reflector-client neighbor MPLS-SP-SPOKES next-hop-self all neighbor MPLS-SP-SPOKES weight 50000 neighbor MPLS-SP-SPOKES soft-reconfiguration inbound neighbor MPLS-SP-SPOKES route-map MPLS-SP-IN in neighbor MPLS-SP-SPOKES route-map MPLS-SP-OUT out distance bgp 201 19 200 exit-address-family
Hello @thefilmguy ,
>> What's the best way to get the MPLS (DMVPN Headend Routers) to share BGP between each other if all routers in the topology are in the same AS?
In addition to have a standard iBGP session between them the two headend routers should act as route reflector servers for all the other spoke routers.
router bgp 6500
! additional commands to make the spokes to be seen as clients:
neighbor SITEA route-reflector-client
neighbor SITE B route-reflector-client
neighbor <other-HE> remote-as 6500
explanation iBGP uses a split horizon rule so that a router RB cannot advertise in iBGP to RC what has learned in iBGP from RA. This rule would mean that a full mesh of iBGP sessions between all routers is needed to ensure end to end propagation of routes. BGP RRS are a tool that allows to save on the number of total iBGP session needed. They are allowed to reflect = propagate routes because they add two BGP attributes to the prefix the BGP originator = BGP RID of device that injected the prefix in the iBGP domain and cluster list that is a list of all the reflections the advertisement has gone through.
In this way reflection is safe on RRS.
Hope to help
MPLS to MPLS is done by SP not from your side.
Inter AS MPLS is done by
Now depend on both SP you can select which one is best for you and for them.
2) MPLS service providers that are not sharing BGP.
Is it possible those two MPLS rtrs are already the RR for the ASN as such there isn’t any requirement for a full ibgp mesh