Re: Disk Percent Busy, out of whack??

Once upon a time I had a calculation for what was used behind
the scenes to come up with average numbers for all drives in an
ASP. This basic calculation was modified for the each unit as
seen in WRKDSKSTS. Unfortunately, I can't find the specific
information I'm thinking of but here is some information as food
for thought .....

Once upon a time Disk utilization = Disk busy = Service time *
System operations / number of disks

which has to be modified a number of times to incorporate things
like RAID protection and caches so in the end you really can no
longer tell device busy.

I think what could be happening is the caching of writes in the
IOA is better on all the drives except unit 1 in your specific
situation. Why unit 1 isn't getting good cache hits, unknown.

Or you may have uncovered some sort of a bug in the calculation.

DrFranken <midrange@xxxxxxxxxxxx> wrote on Wed, 01 Oct 2014
01:53:57 GMT:

I watched it for some time and even for 5 minutes it was over
30% for the one drive and 1% for the others.

I reset at 5 minutes and watched for several more observing
the same activity.

Once the CRTNWSSTG was completed and disk I/O fell to zero
then when I started using the iASP then it 'seemed' normal.

I do have the navigator monitors running and will further
investigate on THursday when I am on site there again.

I think the biggest question in my head is how is that % Busy
calculated? Clearly it's not just reads and writes counters
times the size of the reads and writes blocks divided into
some theoretical maximum I/O for the type of drive. And this
itself isn't surprising. I'm guessing that seek distance and
write pattern contribute to the drive busyness. Just doesn't
seem like it would be THAT different!