We have two servers (HP9000) configured in an MCSG HA cluster. They both have appropriate licenses and both can support full set of interfaces. We have first node running our LIVE environment consisting of 3 (three) sites with over 200 threads. The second server supports TEST and TRAIN environments with total of 6 (six) sites and about 350+ threads. I fully expect each one of the servers to run all nine sites in case of fail over event. As a matter of fact I had one of them do just that when I was recently upgrading from 3.8.1P to 5.3.
