Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: compute statistics for a big table for the first time

Re: compute statistics for a big table for the first time

From: Sybrand Bakker <gooiditweg_at_nospam.demon.nl>
Date: Thu, 27 Feb 2003 21:06:57 +0100
Message-ID: <jrrs5vknanf3sfc3uimph7mmi61i444mj4@4ax.com>


On Thu, 27 Feb 2003 12:16:42 +0100, cris <cris119_at_operamail.com> wrote:

>I wish to compute statistics for a big table (more than one million
>record) that was never analyzed.
>I'm wondering if a single 'ANALYZE TABLE ... COMPUTE STATISTICS'
>statement is too heavy for my server and if it's better to analyze a
>single partition at a time (the table is partitioned).
>Any hints would be appreciated.
>
>Platform: SuSE Linux SLES7
>Oracle version: 9.0.1.3
>Hardware: Pentium III 996.894 MHz 878 MB RAM, RAID5
>
>Kind regards, Cris

A few remarks
- 1 Million records is definitely not much nowadays - ANALYZE should be considered obsolete, you should move to dbms_stats, which also seem to pose fewer problems to resource consumption
- IIRC the most frequently occurring problem during ANALYZE is running out of TEMP tablespace.

Sybrand Bakker, Senior Oracle DBA

To reply remove -verwijderdit from my e-mail address Received on Thu Feb 27 2003 - 14:06:57 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US