physical disk :Avg.Disk queue length

Hi !
I am collecting this counter as well as others using
performence monitor

I know that physical disk:avg disk queue length never should be bigger then 2 per disk
Now i am wondering if my data (mdf,ldf) is being kept only on one drive
Do i have to tace into considerartion others?

i have 4 drives

and sometimes avb.disk queue(total) goes up to 8 is that something to invastigate?
or this is okey?

the average disk queue should be less than 2 per actual disk, not necessarily never, in fact it can burst as high as several hundred during checkpoints,

I am not inclined to set a hard number on either the queue or duration, on the thinking that it is really a judgment call. sql server may dump a load of disk IOs, so there is no easy way to guarantee the queue will be less than 2/disk.
It really is the ability of the disk system to process the outstanding requests quickly. If the queue jumps, one would like to see a quick clearing of the queue

Thank you but actually i was asking if i have to look at all drives or only on that holding the data (mdf,ldf)?

THANK YOU

I am also not sure if this is okey:

I collected physicaldisk:%disk time during couple of days and notice that valued there would be like this during like 20 min at the middle of the day

35.15444699
346.8391533
732.8421681
2.64263303
2.907129939
2.302014733
5.690481086
507.1650232
3.102909192
3.736353902
3.195794051
817.3138379
1.37064159
633.3904984
2.182851514
1.978517605
1151.305497
1.695320746
1.336838025
542.2080446
12.99697185
228.0945295
4.549434747
411.8101916
208.6255572
1078.658681
1404.970601
2.777355502
3.568689506
2.690905888
1.591463755
614.0977082
2.198081054
2.796216522
5.446923527
3.57819483
765.4298031
326.2525327
2.705199415
3.1717983
3.309912987
1.571566058
67.33584523
1685.572359
666.2808927

those values goes like 3-4 per minnutes.
I know that if %disk time is constantly high during 10 min than it's require invastigation
What about those counters ?Are they point to the problem

Also if you are looking at the _Total instance, then you may want to split it into the actual disks (although your original post suggests that you may only have one disk)... This way you may be able to ignore certain known activities such as performance monitoring, profiling, paging, etc.

Also check that pages/sec is close to 0 on average. If not, then although your disks are busy, it is due to memory shortages.

Cheers
Twan

Memory<img src='/community/emoticons/emotion-4.gif' alt='' />ages/sec =average=10.09923107 <br />I collected those data for a day<br /><br />Also i realy have 4 drives but only one hold mdf and ldf files.<br />so i will monitor disk that hold the db files and come back here <br />

i Understand that this is too short periond of time for making conclusions
but after monitoring for 30 min i noticed that when disk writes goes to 200-300
the Curent disk queue lenth remains =0,sometimes it jumps to 4

but once %disk READ time jumped to 400 and in that moment curent disk queue lentth was =37

memory pages/sec seems high...

can you look at the SQL counters buffer page reads/writes? If these are also averaging 10/sec then it sounds like a memory shortage rather than a disk constaint

also check SQLServer Memory Manager total and target server memory. These should be roughly the same

another one is SQL counter buffer Page life expectancy. If this is low and or fluctuates a lot, then consider running profiler to look for long running queries that might be using large table/clustered index scans, flushing out SQL's cache

What else are you running on the server apart from SQL Server? If you are running COM, IIS, file serving, etc. then consider setting SQL Server's max memory to leave 50-100 MB for other things...?

Cheers
Twan

It is fairly rare to have a disk problem, usually the problems are application or memory related, but they appear like disk problems

thank you i will cath them tomorrow.

Today monitoring shows that
SQLServer:Buffer ManagerPage life expectancy=
max=36 614
min=54
avg=16 154
where majority= near or above 30 000

then for acouple a min = 1 600 starting from 54


SQLServer:Buffer ManagerPage writes/sec
avg=9.21
majority=0
couple of spikes of 200 close to each other


SQLServer:Buffer ManagerPage reads/sec
avg=23.7
lot's of 0
SQLServer:Memory ManagerTarget Server Memory always exactly the same like
SQLServer:Memory ManagerTotal Server Memory


sql has 4 processors
physical memory=2560
virtual memory=4000-4000
Microsoft SQL Server 2000 - 8.00.760

Do page reads + page writes /sec follow a similar pattern to memory pages/sec? If not then SQL is not the process causing the paging

Cheers
Twan