Re: parallel query

> but the worrying thing is the 2 sorts in the disk.

Why does that worry your the most? Have you determined where you query is spending the most of its time waiting? If your query is spending 75% of its time performing disk reads, 22% of its time consuming CPU, and 3% of its time performing sorts, then tuning the sorts will give you the least benefit. Put it another way....if your query spent 4 minutes processing, then with the percentages above, you would be spending 3 minutes performing disk reads, .12 minutes performing the sorts, and the balance on CPU. Even if you eliminated the sorts, you would reduce your processing from 4 minutes down to 3.88 minutes. Is that going to be noticable?

Please focus your tuning efforts on the area that will bring you the biggest benefit. Now I just made up those percentages, but you will want to perform a similar analysis yourself. See where you are spending the most time waiting and tune there.

 From the information you have given, I would suspect that tuning your physical reads would bring a bigger benefit than tuning sorts. But I add that with the caveat that I have limited information and do not know where you are spending your time waiting. You mentioned that you are joining 3 tables with 7 million rows each. Have you tried different join methods? Have you ensured that your stats are up to data on each object? Have you looked at partitioning? These are just a few things to look at, in addition to the many suggestions already given.

HTH,
Brian

-- 
===================================================================

Brian Peasland
dba_at_nospam.peasland.net
http://www.peasland.net

Remove the "nospam." from the email address to email me.


"I can give it to you cheap, quick, and good.
Now pick two out of the three" - Unknown