I have this scenario and looking for best practice or suggestion on how to convert a production ospf backbone area 0 to another area with minimal traffic distruption (if that is even possible). I sort of have an idea but want some feedback on the correct/better approach.
Currently there are two separate networks X and Y. The core router in each is configured as ospf area 0. Routers X and Y are to be ospf peers and X's subnets along with the existing devices it is peering with should be moved to a different area, say area 3.
My idea is to create a new ospf process (ospf process 2) on router X and the rest of its existing peers. Router X would peer with Y in the backbone area on the new ospf process (so area 0 would exist on ospf process 1 and process 2). X would also then peer with its existing peers in new area 3 in process 2. Then its a matter of migrating the existing ospf interfaces from area 0 to area 3 from one ospf process to the other. Would this work, or is there a better solution for this kind of migration (a document/whitepaper would be ideal)? Diagram below:
Hi Reza, thanks for your reply. So my understanding on your suggestion is to extend ospf area 0 between X and Y. So this will create a merged area 0 between A, Y, X and B. Next when you bring up a second link between X and B, configuring peering and out them in area 3 in a different ospf process. However after unplugging the first link, then wouldn't there be a partition as area 3 would not have a backbone area? On X, area 0 in is one process and area 3 in another (ospf databases area separated).
