April 25, 2008 at 9:19 am
Hi,
Please could I have some advice I am running a query on the live node of a cluster it takes well over an hour to run even at a quite ish time when CPU is bellow 50% memory disk queues all normal. But when I run it the other node it take less than a minute.
I know it would probably take longer on the live due to having users on but not that much longer, there is no blocking why is there such a huge differences.
Many thanks
April 25, 2008 at 9:31 am
Do the two databases have the same amount of data on them? Do you rebuild your indexes and update your statistics regularly? Have you compared the execution plans on each server?
John
April 25, 2008 at 9:34 am
I don't have any experience with clusters, but it sounds like to each node has its own plan cached for the query. Try executing it on the slower node with the RECOMPILE option.
Jack Corbett
Consultant - Straight Path Solutions
Check out these links on how to get faster and more accurate answers:
Forum Etiquette: How to post data/code on a forum to get the best help
Need an Answer? Actually, No ... You Need a Question
April 25, 2008 at 9:55 am
Edward (4/25/2008)
Hi,Please could I have some advice I am running a query on the live node of a cluster it takes well over an hour to run even at a quite ish time when CPU is bellow 50% memory disk queues all normal. But when I run it the other node it take less than a minute.
I know it would probably take longer on the live due to having users on but not that much longer, there is no blocking why is there such a huge differences.
Many thanks
What clustering technology are you using? A windows cluster would be active/passive, so the only node "seeing" the databases and running the database service would be the "active" one (unless you have some kind of mirroring setup, etc...)
----------------------------------------------------------------------------------
Your lack of planning does not constitute an emergency on my part...unless you're my manager...or a director and above...or a really loud-spoken end-user..All right - what was my emergency again?
April 25, 2008 at 3:31 pm
Please post the query plans for both.
[font="Times New Roman"]-- RBarryYoung[/font], [font="Times New Roman"] (302)375-0451[/font] blog: MovingSQL.com, Twitter: @RBarryYoung[font="Arial Black"]
Proactive Performance Solutions, Inc. [/font][font="Verdana"] "Performance is our middle name."[/font]
April 25, 2008 at 5:33 pm
Keep in mind that a cluster shares the disk, so it shouldn't be IO, however you should be sure that your connections (SCSI or SAN fabrics) are set the same. I'd check that early to be sure that you're not chasing something else.
The two machines are running two separate instances of SQL Server and essentially hand off the shared pieces (databases, jobs, etc.). I wouldn't think that different server settings would be allowed, but I supposed if you connected to one of the instances (not the virtual), you might be able to change something.
Does it work fast if you failover? Is that how you're testing?
Follow me on Twitter: http://www.twitter.com/way0utwest
Forum Etiquette: How to post data/code on a forum to get the best help
My Blog: www.voiceofthedba.com
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply