Re: Analyze gather statistics automated in 10g?
From: Martin Klier <usn_at_usn-it.de>
Date: Tue, 31 Mar 2009 16:48:47 +0200
Message-ID: <49D22D4F.9060104_at_usn-it.de>
Hi Kumar!
> Can you explain why this may happen because we are collecting stats
> and why would one table have 'old' stats. It could be an intermediate
> stage during the stats collection process but would it remain like
> that for the entire day ?
It IS indeed an intermediate state. The stats are collected object wise, but depending on the complexity of your statements it's possible that you are using a table with and a table without new statistics. This happens, if a hard parse occurs during stats gathering.
Date: Tue, 31 Mar 2009 16:48:47 +0200
Message-ID: <49D22D4F.9060104_at_usn-it.de>
Hi Kumar!
Kumar Madduri schrieb:
> . One table already
>> has the new stats, and the other table still has its old stats
> Can you explain why this may happen because we are collecting stats
> and why would one table have 'old' stats. It could be an intermediate
> stage during the stats collection process but would it remain like
> that for the entire day ?
It IS indeed an intermediate state. The stats are collected object wise, but depending on the complexity of your statements it's possible that you are using a table with and a table without new statistics. This happens, if a hard parse occurs during stats gathering.
But if you are executing this staement frequently now, it won't be hard parsed again. You are using an execution plan based on old AND awry stats now! In this situation, you need one of the several methods to force a hard parse once. It it's newl parsed now, all is well again.
HTH
Martin Klier
-- Usn's IT Blog for Linux, Oracle, Asterisk http://www.usn-it.de -- http://www.freelists.org/webpage/oracle-lReceived on Tue Mar 31 2009 - 09:48:47 CDT