Re: Oracle AWR - Time Model Statistics / background elapsed time < background cpu time

From: Stefan Koehler <contact_at_soocs.de>
Date: Fri, 13 Mar 2015 19:35:47 +0100 (CET)
Message-ID: <1830980868.217615.1426271747623.JavaMail.open-xchange_at_app01.ox.hosteurope.de>

Hi Mark,
the main difference is that the "background elapsed time" includes the "background cpu time", but my numbers from the AWR report are not consistent. The "Time Model Statistics" are hierarchical. Once again my numbers from AWR report for verification:

AWR Elapsed Time ≈ 3600 seconds
Amount of CPUs = 160
background cpu time ≈ 47342 seconds
background elapsed time ≈ 33676 seconds

By the way Joel Goodman also mentions this in his blog post( https://dbatrain.wordpress.com/2010/11/22/do-you-have-an-oracle-background/ ): "Using this allows us to see what proportion of time is used by the backgrounds in a hierarchical fashion. The “background elapsed time” statistic includes the “background cpu time” and the “RMAN cpu time” statistics".

In my case the "background cpu time" is much higher than the "background elapsed time", which obviously can not be true.

So my question is "What is causing or can cause this inconsistency here?" as i have found no MOS bug note on this.

Best Regards
Stefan Koehler

Freelance Oracle performance consultant and researcher Homepage: http://www.soocs.de
Twitter: _at_OracleSK

> MARK BRINSMEAD <mark.brinsmead_at_gmail.com> hat am 13. März 2015 um 19:02 geschrieben:
>
> I don't see how "background" or "foreground" makes a difference. There is more than one background process, and there is more than one CPU.
>
> I can't imagine any reason that you should not be able to accumulate background CPU time faster than elapsed time. Have I missed something obvious?

--
http://www.freelists.org/webpage/oracle-l

Received on Fri Mar 13 2015 - 19:35:47 CET