I’m not at all surprised about the login problems, as security is often the link that causes the most problems when trying to hand off between multiple servers. As you pointed out, simply doing a fresh login would probably clear up most of those.

I know you were discussing it before, I don’t know what the outcome was. Load Balancing is an option, but it tends to be hardware heavy, in that you’ll end up with a cluster of servers all generating separate output to individual requests, all managed through the Load Balancer. In other words, if I request “Page A” and someone else requests “Page A”, the page will be completely generated from whichever server is tasked with that particular page for each request.

On the other hand, a reverse proxy can cache the output of the most commonly accessed pages, relieving the backend server(s) tasked with generating new pages relatively unloaded as the request never gets that far for many of the requests made. If I request “Page A” and someone else requests “Page A” within the next few minutes, only the cached page will be sent to the second (third, fourth…) requester, without having to completely regenerate the new page.

You still really need a minimum of two servers (technically, I believe that you can do it with one, but it is more complicated, and nowhere near as efficient) but it doesn’t require as many servers as load balancing does.