We're currently investigating instances when response times spike, as our max times are orders of magnitude different than our average times. Knowing this information would allow us to better speculate as to root causes of these spikes so we can better mitigate them.
Is it possible to learn when these spikes happen?