We have a random period of time when the load pattern of the website causes excessive cpu usage and slowdown for 5-10 minutes. We plan to run the profiler in sampling mode for a day on the live server trying to capture the bogus period. Using other monitoring tools we will know when the problem happens.
After we have a day of profiler data we want to zoom or filter to that given period to check which methods caused the cpu usage.
My question is, is it possible to filter the trace after measuring to zoom into a certain period? I would prefer this in sampling mode in order to avoid slowing down the service.