BIG-IP : configure 2-node pool as simple failover

If you plan to add or think you may need more members in the future, I'd use something else than default round robin (RR). Especially in case of two member pool (in same priority group), RR does not quickly level to even load if one member goes down and comes up again later.

For example if you have 200 connections to the pool, RR ideally has 100 on each. Now member 1 goes down and the member 2 takes the full load (200 connections or whatever number of connections are recreated to it while the other member is down). When the member 1 comes up again it will take the next, i.e. having 1 connection, then the next new connection goes to the member 2 ending up as: member 1:1, member 2:201. Then for the next ones the connections will be 2, 202 ; 3, 203 and so on. Of course some connections will close and so the load typically goes more even, but over long time. I have seen this in practice and used least connections (with some ramp up time) instead of RR.