Re: poor clob performance with very high CPU

From: Ricardo Santos <saints.richard_at_gmail.com>
Date: Mon, 2 Nov 2009 19:49:21 +0000
Message-ID: <34e16fec0911021149p135f0de4wd68630edcdb58a53_at_mail.gmail.com>

Hi Josh,
my first advice to you is to update your Oracle installation to patchset 3 (10.2.0.4.0) and then check out again the metrics of your process, 10.2.0.1 is the base releease of Oracle 10g R2 and is strongly recommended that you update it to one of the lastest patchset, after a fresh installation.

About the storage clauses on your tablespaces for the CLOBS, since we're talking about large objects wouldn't you consider to have uniform extent sizes of a bigger size ?

Best Regards,
Ricardo Santos.

On Mon, Nov 2, 2009 at 7:30 PM, Josh Collier <Josh.Collier_at_banfield.net>wrote:

> Solaris 10 update 4 on SPARC
>
> 10.2.0.1
>
>
>
> All,
>
>
>
> I have a CTAS that involves a CLOB. The expected duration of this process
> is 180 minutes. Recently it started taking > 5 hours. The data and rowcounts
> are very similar, the trace shows only CPU consumption and very little else.
> I am wondering if you guys have any insight into how to diagnose a clob
> performance issue where the only information I have is excessive CPU
> consumption.
>
>
>
> The execution paths have not changed. These are staging tables and they
> never have statistics as we always want full tablescans. The tablespaces
> underlying the tables/clobs have not changed. They are local ASSM with
> uniform extent sizes of 128k.
>
>
>
> Thanks for your time,
>
>
>
> Josh C.
>
>
>

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

Received on Mon Nov 02 2009 - 13:49:21 CST