May 26, 2010 at 1:33 am
Hi everyone,
Yesterday I was having problems with one of our servers where the disk activity was going through the roof (high average disk queue length and Current Disk Queue Length recorded in perfmon). Unfortunately it just so happened to be the disk where the majority of our database data files are and so the systems that rely on these databases were running very slowly.
I was using process explorer to try and find out which processes were generating lots of I/O on the disk but that didn't give me a great deal back. I don't believe it's the database engine playing up, but I was wondering if anyone had any tips on how to monitor disk activity and find out what exactly might be causing the hard drive on the server to go bananas!
So not strictly speaking a SQL 2005 Admin topic, but we DBA's have to be pretty versatile!
Many thanks,
Phil
May 26, 2010 at 7:10 pm
Hi Phil,
Since you used Process Explorer you might want to consider Process Monitor.
The key is to enable filtering for file activity only, configure it to drop filtered events and set the logging to capture to a backing file instead of the page file; start by running with the /noconnect switch. It should only be necessary to run that for a short time to collect enough detail on a busy system.
I just went through this same process on a server that was showing queue lengths of 10-15. The File Summary report showed the files by IO activity, which made it easy to identify the application. It turned out that the applications debugging diagnostics had been left enabled and were causing the most IO. Disabled it and the queue went down to 2.
May 27, 2010 at 9:26 am
Thanks Todd,
I've had a quick look and should be able to get what I'm after. Naturally though the server's started behaving itself so i might never get to the root cause of the problem we were experiencing. Still good tip for the future.
Thanks!
Phil
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply