Wow. That's a lot of resources on a single machine. What kinds of resources are they? Are they running custom monitor scripts or things included with VCS?
It sounds like the issue is pretty reliable. Can you watch what is going on when the CPU spikes and see which monitors are doing it? Maybe there are just a few that are having problems.
Is there any way to stagger the monitoring? I haven't had to do anything like this, but I'm wondering if there is a way to set it so not every monitor runs at the same time. Since you are on a 5 minute cycle, maybe find a way to do a subset every minute. So, at 0 you run the first 60 of them and then each minute after you run a different set so that instead of a huge rush all at once there is a more constant load. Again, I don't know if this is even possible.
