With three routers in a point-to-point row:

[rtr1]<-->[rtr0]<-->[rtr2]

what are the principal differences between these three ospf designs:

A) Three areas: area 0 in the middle, one ospf process/router.

B) It's all area 0, one ospf process/router.

C) No area 0. The endpoints (rtr1, rtr2) have one ospf process each with one area defined, covering the interface between it and the neighbor.

The internal router has two ospf processes defined, each like above, with one area per process and "redist ospf <other-process>"

I note that despite the cisco ospf design guide saying that there has to be a area 0, it seems that is not true in practice (i.e., C works).

Thanks, -mark