February 19, 2010 at 11:25 am
Hi,
We have SQL Server 2005 EE x64 with SP3 and have 1 SQL instance with Share point databases. While taking full backups, I'm getting the High Disk Queue length alarm daily from Spot light as below
Time Connection Action Details Severity Alarm
02/18/2010 11:16:14 PM MOSS1 Alarm raisedDisk Z: has an average queue length of 2.7.HighDisk Queue Length Alarm
Is that normal to get high disk queue length while taking Full backups?
Thanks
February 19, 2010 at 2:23 pm
It depends on how you've laid you files out on your disks. The disk alert references disk Z. What is on Z? Also, keep in mind that 2.7 may not be that bad depending on the RAID configuration for that drive.
February 19, 2010 at 2:38 pm
Thank you,
Disk Z is to keep the back files and we have all drives on SAN with raid 10 configuration
C - OS
D - Data files
E - Log files
T - Tempdb
Z - backups
February 19, 2010 at 2:51 pm
So your Z drive is where your backup files are being written. It is a RAID 10, but how many disks? You may just need to adjust your alert to fit your disk setup. 2.7 is not that bad if your RAID 10 set has 6 disks, this equates to only .45 queue length per disk which is well within the recommended limit of 2 per disk.
February 19, 2010 at 3:20 pm
It is a RAID 10, but how many disks?
Is that mean no.of disks in Z drive? Could you plz tell me where to check that no.of disks for Z drive?
thanks
February 19, 2010 at 3:29 pm
February 19, 2010 at 3:40 pm
DiskDescriptionType Size Device Type Partitions
0Disk driveSCSI50.1 GBSCSI (Port:0, Target ID:0, LUN:0)1
1Disk driveSCSI150 GBSCSI (Port:1, Target ID:0, LUN:1)1
2Disk driveSCSI150 GBSCSI (Port:1, Target ID:0, LUN:2)1
3Disk driveSCSI100 GBSCSI (Port:1, Target ID:0, LUN:3)1
4Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:4)
5Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:5)
6Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:6)
7Disk driveSCSI5.03 GBSCSI (Port:1, Target ID:0, LUN:7)1
8Disk driveSCSI2.01 GBSCSI (Port:1, Target ID:0, LUN:8)1
9Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:9)
10Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:10)
11Disk driveSCSI50.1 GBSCSI (Port:1, Target ID:0, LUN:11)1
12Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:12)
13Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:13)
14Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:14)
15Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:15)
16Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:16)
17Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:17)
18Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:18)
19Disk driveSCSI0.00 KBSCSI (Port:1, Target ID:0, LUN:19)
20Disk driveSCSI100 GBSCSI (Port:1, Target ID:0, LUN:20)1
In above, I can see drives on the node C - 50 GB, D - 150 GB, E - 150 GB, F- 100 GB, Q - 5.03 GB, Y - 2.01 GB, Z - 100 GB, T - 50 GB
I did not understand the 0.00 KB values..are they disks that you referring to?
please advice
February 19, 2010 at 3:45 pm
I'm guess that your monitoring software alerts when your Disk Queue Length hits 2.
The no of disk that make up the array that hosts the z:\ drive need to be considered when evaluating this value.
Also Backups are IO intensive so it is not uncommon to see an increase in IO during backups.
Gethyn Elliswww.gethynellis.com
February 19, 2010 at 4:04 pm
I'm guess that your monitoring software alerts when your Disk Queue Length hits 2.
Yes, the monitoring tool sends an alarm when the Disk queue length is >2
The no of disk that make up the array that hosts the z:\ drive need to be considered when evaluating this value.
I came to know that each drive is created with one LUN. i.e each drive is a single disk, according to our SAN Admin.
So the no of disks that make up the array that hosts the z:\ drive is 1 right? please correct me
and in this case, how can we calculate Disk Queue length?
thanks
February 19, 2010 at 4:13 pm
That not an ideal setup for a production database server.
How would you cope if the one disk that houses your data, or OS fails, there are many articles out there about the IO config for SQL Server...Maybe worth having a chat with your SAN guy.
Gethyn Elliswww.gethynellis.com
February 19, 2010 at 8:18 pm
klnsuddu (2/19/2010)
So the no of disks that make up the array that hosts the z:\ drive is 1 right? please correct me
This does not sound right - there is no way you can have a RAID 10 array with a single drive. So, either you do not have any RAID level setup on that drive - or there are more drives.
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”
― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs
February 20, 2010 at 4:53 pm
I agree. That does not sound right. Either your Z:\ drive is not a RAID 10 (and is in fact a RAID 0 - no RAID) or it is made up of multiple disks.
What you'll need to do is find out which of these is true. If it is truely made up of 1 disk, I'd raise concerns with your SAN admin about redundancy and the possibility of losing not just your backup files, but the ability to backup as well.
If your Z:\ drive is made up of multiple disks, find out how many. Then adjust our disk queue length alert to be 2 * [# of disks in Z:\]. A disk queue length of 2 on a RAID array with 6 disks is not bad. A disk queue length of 12 on a RAID array with 6 disks would begin to worry me.
Viewing 12 posts - 1 through 11 (of 11 total)
You must be logged in to reply to this topic. Login to reply