Asvc_t values in iostat output

UNIX Solaris
1

Noticed that asvc_t values in iostat command outputs are mostly more than 100 in our previous iostat analysis.

Also found the following detail from an alternate site IO Bottleneck - Disk performance issue - UnixArena

----

  1. asvc_t average service time of active transactions, in milliseconds. If the average service time is exceeding more than 25, then there is issue in the specific disk and need attention.
    ----

Please let me know whether the above statement is true?

2

Watch it with

iostat -nx 2

to get the actual values.

asvc_t above 25 ms for a disk is high: consider a faster disk,
or (better) try to optimize your application!
Or add RAM in case there is paging activity with

vmstat 2

For nfs devices an asvc_t above 100ms is still okay.

3

The ascv_t time varies all over the place. A SATA 7200rpm disk will be saturated at a lower asvc_t number than a disk with higher rpms.

The point is that those numbers are relative. You have to work with them under various system conditions to really understand what you are seeing. MadeInGermany is giving a good start, but do not jump onto 'fixing' things until you are sure you understand the problems. It could simply be a poorly written piece of code eating a disk for lunch.

That 25 number is mostly baloney. IMO. I have a SAN with huge IOPS capacity. A number like 24 would indicate an unbelievably high load. I could see that by looking at r/s and w/s for that disk. On some other device, it is okay.

If you need help with tuning, consider sarcheck - it does a good job of sroting through sar data and telling you what is happening. IMO