I understand what you're saying about the loadbalancer perhaps sending the user to another ip however, they handle that by setting a cookie on the user's browser of the IP for the machine it was selected it to use (they call this BIGip). That way it consistently connects to the same server.
I set up a test for this scenario where I built a script that basicly set a session and then reloads the page, counting it each time, until the session was lost. I can put 3 seperate browser windows up, one for each server ip, and start them simultaneously. I was expecting all 3 servers to lose the session at the same time. Instead, two of them will stop at the same time, while one will keep counting, but eventually will stop as well. The funny thing is, its never the same two servers to lose the session at the same time, its random.
Perhaps the easiest test will be to bypass the loadbalancer by taking it off the network for one server and try this test again. If I lose a session then it must be something other than the loadbalancer.
Thanks for your input guys, its really appreciated!