Hello. Recently I’ve been learning in a lab about MPLS infrastructure. I’ve got questions from my lecturer about redundancy in an MPLS backbone. The MPLS L3 VPN lesson is already good enough start. I’ve tried to improvise from the lesson and did some configuration, managed to enable redundancy by simply adding new networks to the OSPF process. I need advice in enabling redundancy to the MPLS backbone. I’ve attached the topology below. Thanks 
Hello Evan,
Let’s start with the physical topology first:
- Your customerA-HQ and customerB-HQ routers are attached to a single PE router. Same thing fo the branch routers but your HQ routers are probably a bit more important. You could connect these to two PE routers.
- Your PE routers are connected to a single core router.
Take a look at this topology:
This is a nice example where a single router is connected to two PE routers and we have redundancy within the service provider network. Each PE router is connected to two core routers.
Rene
Thanks for the answer Rene, Now that I’ve done that with the physical topology. What other strategies should I configure so that it has redundancy on each PE routers?
Hi Evan,
With a topology like this, you have a fully redundant service provider network. On each side, you have two PE routers so a single PE failure is no problem… In your core, it’s no problem when a core router fails. Your IGP in the SP network will route around failures.
You also might want to think about MP-BGP…You could create a full mesh of iBGP peerings with PE1/PE2/PE3/PE4. This is no problem but not a very scalable solution. Instead, you could use a route reflector but that means your RR will be a single point of failure. You can add a second one…
On the customer side, there is no redundancy so you could add a second CE router for each site.
There are some options to speed up convergence when there are failures in your network. Things like OSPF FRR, BGP PIC, etc.
Rene