How does this work?

It's a novel solution and should provide very quick "persistence" to the two pools, but it doesn't provide any kind of higher level functionality that you would get if you added all the members to the same pool and used a load balancing algorithm like least connections or observed to select the best pool member. In v9 or 10 it also doesn't offer any persistence within the two different pools (persistence was set at the pool level in 4.x so this wouldn't have been an issue).

 

 

I'd say you'd be better off changing the pool configuration to one pool, use persistence on the virtual server and a more advanced load balancing algorithm. You can read sections of the LTM config guide for your version to get details on the various load balancing and persistence options.

 

 

Aaron