"My only objection to your solution was to express my opinion . . ."
Ah, but I never offered a solution, I suggested ". . . you might investigate whether something like Cisco's PfR might be suitable for your goals."
Heck, notice I wrote "like Cisco's PfR", i.e. the concept, not necessarily the actual technology or product.
Also, I started my posting with the question: "What is it you're really trying to accomplish?", because it was unclear, to me, whether OSPF was a "must", or are the end-goals a bit broader.
Anyway, you imply you did look at/into PfR, and believe:
- It is impossible to have static routing for over 1k branches.
- There are many complexities in the PfR configuration
- and I think it is an outdated solution
I fully agree with there can be many complexities in the PfR configuration. Unsure how/why your comment on static routing vs. over 1K branches is relevant to PfR. Or, why you believe it's outdated; especially when compared to doing dynamic rerouting with OSPF (somehow).
BTW, in the past, I've actually used PfR, and, it was "magic"; doing things that dynamic routing protocols, like OSPF, could not.
For example, in the case where there is "cloud" congestion, to your site, via one connection, how do you "see" this? What do you do about it? PfR "sees" the congestion, and, assuming there's another path to same site, not as congested, it reroutes flows to the less congested cloud egress.
Laugh, in fact, when I first enabled it, in a production environment, the only "complaint" I got was the network monitoring of possible "cloud" performance issues stopped showing such issues because PfR routed around such problems before they could be detected (or the "what's wrong with the network" phone calls started). (Complexity, yes, we had to figure out how to do network monitoring not "fixed" by PfR and/or analyze PfR logs, why it made a change [so we could hit up our WAN providers for not meeting SLAs, etc.].)
But perhaps I digress too much on PfR itself.
I also agree with you that complexity can be bad, although I'm more against needless complexity, because sometimes complex issues/problems require complex solutions. So, when I read about what you're trying to accomplish using OSPF, I do wonder, if you're too focused on OSPF causing you to not fully consider other approaches.
Lastly, it's your network. If you want to dice and slice OSPF, have at it. I.e. don't let my contrary opinions stop you.
I really did mean "Do let us know how it all works out." because I like to keep aware of novel solutions so I can add them to my bag of tricks.
