Our FVP DB/CI has high CPU spikes in the 2000 hour on Saturdays. Since the SAP-related CPU reported in ST03 is negligible on FVP, I assumed
the high CPU was driven by Oracle. But when I looked at the 2000-2100 hour on Saturday, July 20, I found that some of the AWR data didn't
appear to make sense, given the CPU data. Details are in the attachment.
Basically, AWR seems to be giving inconsistent data on how much of the CPU is being used for database work. Is the strange result I'm seeing in the AWR report due to a hourly interval being insufficiently granular for what I'm measuring? Is there any way to get more granular performance data from Oracle? Or am I just interpreting the report incorrectly?
Thanks very much.