Please start any new threads on our new site at https://forums.sqlteam.com. We've got lots of great SQL Server experts to answer whatever question you can come up with.

 All Forums
 SQL Server 2000 Forums
 SQL Server Administration (2000)
 Perfmon Avg. Disk Queue Length

Author  Topic 

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-08 : 09:02:55
In perfmon for a particular SQL server I am getting an average disk queue length of over 400 thousand! I have read that a value over 2 is bad. Another server running the same application has a value of 0.005!

Has anyone seen anything like this before?!?!

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-08 : 09:31:20
it just hit 10 million!!!!
Go to Top of Page

Michael Valentine Jones
Yak DBA Kernel (pronounced Colonel)

7020 Posts

Posted - 2007-06-08 : 09:34:29
That doesn't even sound possible.


CODO ERGO SUM
Go to Top of Page

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-08 : 09:55:04
The database is on a SAN I am now reading these counters may not be acurate for a SAN?
Go to Top of Page

bpgupta
Yak Posting Veteran

75 Posts

Posted - 2007-06-08 : 09:58:23
You should put this counter for monitor for atleat 1 day and find out the average of all the values .
Go to Top of Page

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-08 : 10:04:47
Go to Top of Page

eyechart
Master Smack Fu Yak Hacker

3575 Posts

Posted - 2007-06-08 : 10:21:53
i never monitor average disk queue length, just current disk queue length. make sure you select this counter for all your drives.


-ec
Go to Top of Page

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-08 : 11:11:54
So I shouldn't worry about the strange values above?
Go to Top of Page

eyechart
Master Smack Fu Yak Hacker

3575 Posts

Posted - 2007-06-08 : 11:32:24
quote:
Originally posted by cornall

So I shouldn't worry about the strange values above?



i didn't mean that :)

I think the average queue length is misleading. Current disk queue length is a better counter imho. Create a counter log and monitor it for several hours to see what the trend is. Make sure you monitor all drive letters in your system btw. If the value is still high, then you need to investigate why your disks are so busy and see if that can be remedied (expensive queries, bad indexes, etc.). If you cannot fix that, you will want to make changes to your disk subsystem so it can support more disk IOs (add spindles, upgrade RAID controller, change to RAID 10, etc.)



-ec
Go to Top of Page

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-08 : 12:26:07
Cheers I will run long term monitoring on the io counters. I am also planning to run SQLIO and SQLIOStress to test the SAN make sure it is getting decent IO performance compared to the same DB running on a SCSI raid config.
Go to Top of Page

rmiao
Master Smack Fu Yak Hacker

7266 Posts

Posted - 2007-06-08 : 13:12:07
Can also let SAN guys monitor channel utilization.
Go to Top of Page

eyechart
Master Smack Fu Yak Hacker

3575 Posts

Posted - 2007-06-08 : 14:19:05
quote:
Originally posted by cornall

Cheers I will run long term monitoring on the io counters. I am also planning to run SQLIO and SQLIOStress to test the SAN make sure it is getting decent IO performance compared to the same DB running on a SCSI raid config.



What SAN vendor are you using? What HBA? What switch? It is also important to check HBA firmware, fiber switch firmware, HBA drivers and such.



-ec
Go to Top of Page

cornall
Posting Yak Master

148 Posts

Posted - 2007-06-11 : 05:34:25
I don't know which SAN at the moment. I am working for an application vendor our application is performing poorly. I suspect it is the disk set up but have to be politicaly sensative about what I say! need to run some monitoring and tests before I can start accussing their SAN!
Go to Top of Page

eyechart
Master Smack Fu Yak Hacker

3575 Posts

Posted - 2007-06-11 : 11:44:15
quote:
Originally posted by cornall

I don't know which SAN at the moment. I am working for an application vendor our application is performing poorly. I suspect it is the disk set up but have to be politicaly sensative about what I say! need to run some monitoring and tests before I can start accussing their SAN!




I would just ask the hardware guys if you are downlevel on hba/switch firmware and drivers. Pretty simple question - and if they get bent out of shape about that question then they are officially whiney bitches. In that case it is legal and appropriate for you to kick sand in their faces and steal their lunch money.



-ec
Go to Top of Page
   

- Advertisement -