RE: Not sure why parallel hint didn't work correctly

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Tue, 16 Feb 2016 22:30:16 +0000
Message-ID: <CE70217733273F49A8A162EE074F64D9282C28F7_at_EXMBX01.thus.corp>


The backup shouldn't have the effect described - you should still see all the servers (in each server set) doing similar amounts of work, not everything happening in one server. (That sounds more like Stefan's suggestion of a skew sending all the data to one server by accident, but that then suggests you've got a very unlucky hash/hash distribution).

The parallel 48 query being more expensive than unhinted with with a table degree of 32 sounds odd, but a parallel(N) hint enables some of the parallel automatic policy stuff, which can do different arithmetic from the manual stuff. Are the "small" tables also declared parallel ? If so, what degree ?

I think we need to see the plan and the query.

Regards
Jonathan Lewis
http://jonathanlewis.wordpress.com
_at_jloracle



From: Sandra Becker [sbecker6925_at_gmail.com] Sent: 16 February 2016 21:01
To: Jonathan Lewis
Cc: oracle-l
Subject: Re: Not sure why parallel hint didn't work correctly

I did forget to mention that an RMAN backup was running and is still running. It's a 30T data warehouse so backups take quite awhile. They have only 4 threads for the backup. They are working on a new backup solution so hopefully we'll get that taken care of in the next 6 weeks. Could the backup play a part in this?

--
http://www.freelists.org/webpage/oracle-l
Received on Tue Feb 16 2016 - 23:30:16 CET

Original text of this message