Re: The best CPU usage measurement in Oracle: BUFFER_GETS or CPU_TIME?
- From: "Binley Lim" <Binley.Lim@xxxxxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Tue, 22 Jun 2004 00:57:52 +1200
----- Original Message -----
From: <J.Velikanovs@xxxxxxxx>
> Lets me ask contrariwise question:
> Are there situations when using statspack report (if using) or other
> instance wide performance tool the better way to see TOP SQL-s is sort by
> BUFFER_GETS not CPU_TIME?
>
I think that someone has already answered that "response time" is the far
more important measure. Someone has also responded that you cannot pick
BUFFER_GETS over CPU_TIME or vice versa for a number of reasons.
To look at TOP-SQLs, I would also look at EXECUTIONS and DISK_READS, and
anything that appears in the top-list needs looking at. For eg, I had one
situation where I missed the problem completely because BUFFER_GETS was
nowhere near the top, but it was killing performance because it was
constantly executing selects over a DB-link to check on a status_code that
seldom changes. It was killing performance because it was flooding the
network, not because it was doing a lot of BUFFER_GETS..
So it all comes down to "response time". But what is that creature? No one
can answer that. It depends on too many things, including -- OS vendors
typically try to optimize their kernels depending on what applications
(Oracle included) demand of it. And different OS vendors do different
things. The next OS release could be different, so any single Oracle metric
that works now could no longer work in future.
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
