Active-Active between two datacenters

Not much notice eh ;-)

Firstly, OTV isn't necessary, L2 adjacency is no longer required for network failover (it's UDP unicast).

I'm not sure how you'll achieve your aim where the local servers are concerned, ConfigSync works against you. I assume you want failover to be automatic?

What failover scenarios are you looking at? Priority group activation may be an idea where each pool has the home DC servers set at a higher priority than the remote ones. If the home DC or servers fail, failover occurs and the home DC servers will no longer be available and thus the remote servers will be used.

I don't think that fits the active/active model very well mind.

The ideal would probably be to not run HA, configure each separately and use some other mechanism (GTM, DNS...) to direct traffic between the two.

WLB, thanks for your response. If anything I say doesn't make sense, then it's because I'm 1) brand new to F5/BIG-IP platform and 2) not very smart.

My original implementation thoughts are something like this:

HA and Virtual Server addresses will be in OTV subnets between the two DCs.

Configure SNAT pool for local addresses, with a self/floating IP in that subnet on both F5s.

This would (in my theory) allow the real server IP addresses to remain local, while allowing for failover in the event one F5 was unavailable. Traffic would traverse the fiber link twice per request which obviously isn't ideal, but it would allow for load balancing to continue.

When it comes time to deploy the other services, my plan was to create a second traffic group with the other LTM as the primary, etc.

Does that make sense?

It does but it's far from ideal. If you can live with the trombone then carry on as you would for any active/active configuration.