Avserv is too high

[Deniz Cendere]

Hi all,,
I have a "sar -d" output like below. There is something strange at that output. My avserv mostly equals to 1-8 ms. But sometimes it becomes to huge values.
What can be the problem with it? Avwait is nearly equals to 0.

Thanks

15:40:41 device %busy avque r+w/s blks/s avwait avserv
15:40:42 c0t6d0 1.00 0.50 2 27 0.00 9.20
c2t6d0 1.00 0.50 2 25 0.00 5.17
c19t0d0 30.00 0.50 64 3664 0.00 6.57
c19t0d1 21.50 0.50 66 3864 0.00 4.53
c19t0d2 31.50 0.50 62 3688 0.00 8.14
c18t1d0 15.50 0.50 66 3728 0.00 3.29
c18t1d1 24.00 0.50 66 3840 0.00 5.42
c18t1d2 15.00 0.50 62 3632 0.00 4.01
c4t1d6 0.50 0.50 1 8 0.00 0.34
15:40:44 c0t6d0 3.02 0.50 6 12 0.00 4.65
c2t6d0 2.51 0.50 6 12 0.00 4.58
c19t0d0 57.79 0.50 11 635 0.00 79.57
c19t0d1 51.76 0.50 10 643 0.00 78.47
c19t0d2 57.29 0.50 10 555 0.00 86.42
c18t1d0 50.25 0.50 9 515 0.00 77.65
c18t1d1 46.73 0.50 8 515 0.00 87.12
c18t1d2 57.29 0.50 12 724 0.00 65.37
15:40:46 c0t6d0 1.49 0.50 4 80 0.00 9.04
c2t6d0 1.49 0.50 3 76 0.00 9.06
c19t0d0 72.14 0.50 11 756 0.00 93.79
c19t0d1 62.69 0.50 13 756 0.00 73.54
c19t0d2 90.05 0.50 17 979 0.00 76.54
c18t1d0 29.35 0.50 12 764 0.00 42.91
c18t1d1 35.32 0.50 13 764 0.00 29.81
c18t1d2 23.88 0.50 14 796 0.00 19.02
15:40:48 c0t6d0 2.50 0.50 4 52 0.00 13.10
c2t6d0 2.00 0.50 3 48 0.00 10.96
c19t0d0 79.50 0.50 22 1344 0.00 66.46
c19t0d1 83.50 0.50 23 1288 0.00 58.14
c19t0d2 81.50 0.50 22 1280 0.00 60.66
c18t1d0 14.50 0.50 25 1528 0.00 6.74
c18t1d1 16.00 0.50 26 1512 0.00 6.89
c18t1d2 1.50 0.50 26 1464 0.00 0.97

[Peter Godron]

Deniz,
have a look at this doc:
http://www4.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000082420754

"avwait = Average time (in milliseconds) that transfer requests waited idly on queue for the device

avserv = Average time (in milliseconds) to service each transfer request (includes seek, rotational latency, and data transfer times) for the device.

When average wait (avwait) is greater than average service time (avserv) it indicates the disk can't keep up with the load during that sample . This is considered a bottleneck ."

[Deniz Cendere]

Ok,
But my avwait is zero. My avserv is too high. What is the meaning of this? No process is waiting for I/O but my disks are slow ?
Thanks

[Peter Godron · Accepted solution]

Deniz,
I would interpret your results as a few requests/long duration situtation.
You have few/no requests waiting, but the duration for each one is relatively long.

Why did you look at the disks, are you having performance problems?

[Deniz Cendere]

Hi,

Can you please explain it more detailed? How does it "sar" calculate this?

Thanks

Deniz,

[Peter Godron]

Deniz,
I don't have access to sar source code, so can not tell you how it calculates the stats.
Have you read "man 1m sa1" for more technical details of sar?

I normally use sar for a quick check only.

For further tools,please see:
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=342346
Or
http://www.sarcheck.com/schp.htm