Viewing 15 posts - 1 through 15 (of 29 total)
Yep.
Anyway, we have not used NOLOCK randomly, it is explicitly used where the system has been designed in a way where dirty reads is either okay or will not...
December 16, 2009 at 8:09 am
After doing some reading on the issue it seems that both missing rows and counting rows twice is connected to page splits.
As we have a clustered index on a...
December 16, 2009 at 6:46 am
GilaMonster (12/16/2009)
one-1016367 (12/16/2009)
I wasn't aware that I could get the same row more than once, in which scenario could that happen?
The link I gave you when I first...
December 16, 2009 at 6:07 am
GilaMonster (12/16/2009)
one-1016367 (12/16/2009)
Yes, we are confident using nolock in the given scenario.
You are aware that there is a chance it can can cause rows to be returned more than once,...
December 16, 2009 at 5:43 am
And just to clarify - in my opinion we have satisfactory performance on this table at the moment, the only issue was this particular SP that suddenly wasn't using the...
December 16, 2009 at 5:18 am
ALZDBA (12/16/2009)
How well organized is your clustering index ? (cluster ratio)
- has your index (table) been rebuild lately ?
- Did you load the...
December 16, 2009 at 4:59 am
Oh, didn't see your other notes:
The date is restricted to maximum one month.
T2.Id is indexed.
The date on T1 is the clustered index.
December 16, 2009 at 2:22 am
No, as I said I haven't written this SP, nor have I attempted to optimize it.
When indexes are used the original SP runs relatively fast, but your example...
December 16, 2009 at 2:20 am
Here's the stored procedure. (We do not do SELECT * , but I have removed/modified the column names to shorten the sp and avoid exposure (as this is a...
December 16, 2009 at 2:01 am
Nope, no table variables. I'm not at work at the moment, and I didn't write the stored procedure myself so I don't remember the details. I'll paste the code...
December 15, 2009 at 11:57 am
Okay, I'll give it some thought. The stored procedure lists transactions in the system by date and merchant - and depending on the size of the merchant the result can...
December 15, 2009 at 10:48 am
GilaMonster (12/15/2009)
Or, if you'd like some specific and...
December 15, 2009 at 7:00 am
GilaMonster (12/15/2009)
It's one of the solutions. I referred you several posts ago to a 3-part series that I wrote on parameter sniffing - http://sqlinthewild.co.za/index.php/2007/11/27/parameter-sniffing/
Sorry, I missed that in your previous...
December 15, 2009 at 4:24 am
So, should I do something along the lines of this?
http://elegantcode.com/2008/05/17/sql-parameter-sniffing-and-what-to-do-about-it/
December 15, 2009 at 2:25 am
Today the SP runs fast and is yet again using the indexes, so it appears that the sp_statistics that were run tonight did the trick.
So I guess it was...
December 15, 2009 at 1:25 am
Viewing 15 posts - 1 through 15 (of 29 total)