Under load, we've noticed response times from Apache vary greatly for the same 7k image. It can range anywhere from .01 seconds to 25 seconds or greater. Unfortunately, due to corporate policy constraints we are pretty much stuck on Apache 2.0.52. I'm at best an Apache novice so I'm in over my head with this problem.
My focus recently has turned to our choice of MPM modules. We use the worker model on a dual core hyper threaded blade. It doesn't appear that swapping is an issue, and I don't see any signs of a hardware problem. I've read that worker is optimal on hardware with many CPU's where prefork it more suitable for our specific hardware profile. I can see conceptually how choosing the wrong MPM could result in this erratic behavior, but I'm not confident that it's the root cause here.
Has anyone else seen this type of range in your response times for simple static content? What else should I be looking into here?