MP-iBGP and Route Reflector Behaviours

[kradjesh13]

Hi,

Let me explain about the design before getting into the details. We are a service provider and we have presence across different data centres located across the country. Our core boxes are the mixture of Cisco VXR’s and 6513 switches which have MP-iBGP peering with the route reflectors. If a new client comes on board, a new VRF will be created to carry the client’s traffic and as a standard practice the VRF will be added to all the core devices across the network even if they don’t have a presence in a data centre.

Now, I am designing a network for a client who has presence only at two fixed locations as shown in the attached diagram. We will be acting as a transit network between the client and another service provider. So, I have planned to use MP-eBGP between us and the other provider and default/static routes to the client’s network I don’t think will be an issue. Since the client has presence only at two locations, my design thoughts are to create a VRF and a Vlan and form an iBGP session only between the two routers and bypassing Route Reflectors. Created the VRF at BNE_R1 and formed MP-eBGP relation with the other provider and I can see some routes appearing via the peering which is normal and expected.

My problem is (not problem I don’t understand how this is happening), I created the VRF on Mel_R1 router and did not add any extra lines of configurations to BGP under that VRF instance and when I was checking some thing I accidentally found the same routes which appears on Bne_R1 is appearing in the VRF’s routing table via MP-iBGP session through route reflector. I can’t understand how this is happening, since I haven’t added any thing on the route reflector and some how its leaking traffic.

1. Is this normal??
   
2. Is it’s a must / standard / Good Practise to add the newly created VRF across all the device which peers with the route reflector ??
   
3. Is there a way to override the Route Reflector just for this client (VRF) and form a direct MP-iBGP peering directly with the devices involved.
   
4. If there is a way for the above point. Is it’s a good practice to bypass route reflector??
   

A network can be designed in much number of ways. But I am looking for the best options and some valuable suggestions from you guys.

Thanks

[Sudeep Valengattil]

Hi,

To answer you,

1. The ebgp peering with the provider cloud  would automatically get redistributed into vpnv4 and would be shared with RR.  And RR inturn would share it with Mel-R1. This is normal and expected.

2. It would not be a good practise, unncessarily creating VRFs on all routers

3. Just stop importing route-target (no route-target <>:<>). It would not use RR then, instead it will use only the prefix learned from VRF ibgp neighbor through vlan 432.

4. Its not a good practise to bypass RR, especially in SP environment. Having RR in network would not cause any harm.

Thanks,

Sudeep