Re: Verify Outdated Optimizer Stats
From: joel garry <joel-garry_at_home.com>
Date: Tue, 17 Feb 2009 13:35:49 -0800 (PST)
Message-ID: <427e2698-de83-4306-906e-f5d883debdcb_at_b8g2000pre.googlegroups.com>
On Feb 16, 8:55 pm, Ram <krishna..._at_gmail.com> wrote:
> Hi All,
>
> I hope every one would have come across that users complaining about
> the response of database.Usually the first thought comes to the DBA
> is to Gather Stats and verify the performance.Performance will be far
> better after collecting stats and this is expected behavior.
>
> Before users start complaining about the performance of database is
> very slow , Can't we verify the statistics for out database is that
> outdated. Can any one throw some light on this .
>
> Regards
>
> RAM.
Other answers are stored outlines and profiles. Check this out: http://jonathanlewis.wordpress.com/2007/02/11/profiles/
Date: Tue, 17 Feb 2009 13:35:49 -0800 (PST)
Message-ID: <427e2698-de83-4306-906e-f5d883debdcb_at_b8g2000pre.googlegroups.com>
On Feb 16, 8:55 pm, Ram <krishna..._at_gmail.com> wrote:
> Hi All,
>
> I hope every one would have come across that users complaining about
> the response of database.Usually the first thought comes to the DBA
> is to Gather Stats and verify the performance.Performance will be far
> better after collecting stats and this is expected behavior.
>
> Before users start complaining about the performance of database is
> very slow , Can't we verify the statistics for out database is that
> outdated. Can any one throw some light on this .
>
> Regards
>
> RAM.
Other answers are stored outlines and profiles. Check this out: http://jonathanlewis.wordpress.com/2007/02/11/profiles/
The first thoughts that come to my mind are "What has changed? What is the scope of the problem? What does a quick look at EM tell me? A quick look at top? Alert log? Can I run the app? Oh sh..."
I seem to be having good luck with the default stats gathering, but some of my big tables have lots of old data that may be distributed differently when imported and statistics were gathered, than now, and the stupid percentage rules mean stats have not been gathered since, where they might be if I archived old data. I haven't yet figured out if this a problem. First thought is, the optimizer doesn't seem to be near any boundary conditions that would change it's chosen path. I'd hate to be unpleasantly surprised, though.
jg
-- _at_home.com is bogus. http://www.nytimes.com/2006/02/19/magazine/19wwln_safire.html?ex=1298005200&en=a5d02adfe7aca7c6&ei=5088&partner=rssnyt&emc=rssReceived on Tue Feb 17 2009 - 15:35:49 CST