Jason Oon, June 20, 2018 - 9:19 am UTC
Thanks for the reply Tom.
I did some random checking on 1 of my production databases, pulled an AWR for 1 whole day & also 1 hour gap. After I did some checking and calculation based on IDLE_TIME & BUSY_TIME values, I found out that the average CPU utlization is only about 12%
The production DB is riding on a 24core CPU machine, does that mean the DB is actually oversized?
June 22, 2018 - 2:27 am UTC
Averages tend to smooth out peaks, so you might need to do something like monitor cpu at an OS level on a more granular level (eg vmstat every 30 seconds or similar).
But if its a legitimate 12% *and* you are monitoring at the right time (ie, at month end you might be busier etc), then its possible you have more headroom then you really require.