RE: Statistics Problem on partitioned table

From: Mark W. Farnham <mwf_at_rsiz.com>
Date: Wed, 2 Oct 2013 08:06:46 -0400
Message-ID: <020401cebf67$df050d30$9d0f2790$_at_rsiz.com>

I fail to understand why the --+ gather_plan_statistics execution of the simple count I requested is not forthcoming.

Now it does leave out the table access for the non-index columns, but it should tell us a great deal about what the optimizer thinks is the cheapest way to get the relevant rowids from all the partitions and the plan should show what the CBO has estimated along with the actuals.

Fragmentary revelation of the underlying facts (obscuring, for example, that a view was involved) does not make it easier for folks to help you explain what you're asking about.

Please try to make it easier.

mwf

-----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mohamed Houri
Sent: Wednesday, October 02, 2013 7:39 AM To: Jonathan Lewis
Cc: ORACLE-L
Subject: Re: Statistics Problem on partitioned table

Jonathan,
I am sorry I may have not been clear but there are in fact 721,699 and that is what I showed above

 *select per_ind, count(1) cnt from XXX_PER_YYY group by per_ind;*

721,699 rows .
<snip>

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

Received on Wed Oct 02 2013 - 14:06:46 CEST