RE: Oracle RAC TEMP / TEMP GROUP tablespace performance question

From: Mark W. Farnham <mwf_at_rsiz.com>
Date: Sat, 16 Feb 2013 09:06:10 -0500
Message-ID: <007e01ce0c4e$c6deb6b0$549c2410$_at_rsiz.com>



Do all your users use the same temp?
Do your users have instance affinity when they log on? Do you have uniform SAME in one piece or separate independent units of I/O (each SAME individually) that are used for specific purposes? Do your tempfiles use storage bandwidth on units of I/O that are also used for regular tablespaces?
Is it okay with you that heavy usage of temp by one user may degrade the performance of another user?

These are input parameters (and others might exist) to answering your question.

mwf

-----Original Message-----

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Christopher.Taylor2_at_parallon.net Sent: Thursday, February 14, 2013 5:02 PM To: oracle-l_at_freelists.org
Subject: Oracle RAC TEMP / TEMP GROUP tablespace performance question

Env: 10.2.0.4 RHEL 5.6 3-node RAC
We have 2 TEMP tablespaces:
TEMP1
TEMP2 We have 1 TEMP GROUP:
TEMP_GRP TEMP1 = 5 tempfiles
TEMP2 = 5 tempfiles

I'm curious if it would make more sense to have multiple TEMP tablepaces, each with 3 datafiles (based on the number of RAC nodes) and 1 TEMP GROUP instead of 2 TEMP tablespaces with multiple tempfiles?

The reason I ask is that I seem to be getting some wonky temp space performance from the current setup and I feel like this could be designed better but I didn't want to start down a path that has been well trod already.

So any of you guys have recommendations on TEMP tablespace setups before I start tearing things apart for testing?

Regards,

Chris Taylor
Oracle DBA
Parallon IT&S

--

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

--

http://www.freelists.org/webpage/oracle-l Received on Sat Feb 16 2013 - 15:06:10 CET

Original text of this message