Re: Performance metrics

From: Karl Arao <>
Date: Thu, 12 Apr 2012 09:09:11 -0500
Message-ID: <>

BTW, aside from putting Adaptive Thresholds on our environment to alert for sudden workload change
We are also running a daily report of whats_changed.sql to keep track of SQL plan changes. This script measures standard deviation (STDDEV) on the elapsed time

and this script can also be modifed to address sudden increase in LIO if the execution in question returns more data than a previous execution (we can also take into consideration the LIO/row stuff) or let's say in Exadata a sudden increase in Smart Scans...

So that's how you get proactive on the SQL side.. but of course the application owners should also keep track of their SQL/Batch run times so they can also alert you if their Batch processes are now slower than the previous runs.


Karl Arao

-- Received on Thu Apr 12 2012 - 09:09:11 CDT

Original text of this message