Some of those HP servers are beasts!! if this is just a typical n-tier, with a LB > web servers > app server > db server with the web servers serving the static images and such, I would expect the db or the app server to take the majority of the load. Depending upon how the scripts were written (what test tool? We use Load Runner almost exclusively except for soa testing), you may have had a huge caching advantage and weren't exercising the system properly. I've saturated t1's with 1k users, depending on what was being served. What was your throughput numbers? Hits/second? Also, how were you load balancing? If you were sticky via IP and you were only coming from 1 or two injector machines, you may not have balanced correctly. To be fair, there are a ton of variables involved here, and I'm not certain this is the forum for it.
However, you've certainly peaked my curiosity (I am a geek :) ) so PM me and I'll be glad to help if I can.
KE
However, you've certainly peaked my curiosity (I am a geek :) ) so PM me and I'll be glad to help if I can.
KE