Hello networking
We run BGP / MPLS L3 VPN in the datacenter with a relatively small number of peers. However, we are going to be increasing the amount of BGP routers, as we replace old equipment that was previously in a "legacy" part of the network. The full mesh IBGP configuration is getting longer and bothersome. I am looking at using router reflectors for scaling, but I am going back and forth between which routers in the topology to use for it. I am somewhat knowledgeable of the rules/limitations of RR in regards to path advertisements and selections, but I'm afraid I'll be negatively surprised by something along the road. The RR would be either a pair of NX7706 or ASR9901 where the NX is in the core and the ASR is outside the dataplane of 90% of the traffic it'd be RR for.
We also run BGP DMVPN with "internal" EBGP design, so we'd not be strangers to using a design such as an AS number for each rack.
I know this post doesn't contain a lot of detail and there are a ton of things to consider, but the implementation is a way down the road. So I'm looking for some general advice or pitfalls that someone has experienced implementing scalable BGP solutions in the datacenter.
No comments:
Post a Comment