RE: Disk arms and faulting: (was - We've Added more memory...but I can't remember!)

Charly,

Thanks for the clear reply.  I do believe you have cleared up my
confusion.  I have a basic understanding of the issues you mentioned.
What I did not understand from your post was the difficulty of measuring
disk activity.  If the queuing multiplier is having a serious effect on
your system, it was my understanding that the actuator activity levels
would provide a measurable indication that the system was exceeding
published guidelines.  It sounds as though your management has expressed
an interest in resolving performance issues.  From your posts, you are
obviously not a slouch in understanding and explaining the causes of
poor performance.

Either using something as simple as the WRKDSKSTS display or some of the
performance reports, I thought that you would be able to demonstrate
activity levels in excess of the 40%-50% that IBM publishes as the knee
of the curve.  Because you didn't mention these high activity levels, I
made the assumption that you had checked and did not have measurable
performance issues with disk activity.  Thus I did not make the
connection between the extra disk accesses from faulting and response
times.

Is this the case?  Are you having difficulty coming up with some
empirical measurement which would convince management of the need to
purchase additional disk?  You've got close to 10 GB of main storage in
your interactive pool, with an activity level of 23 I would imagine that
your system is doing a fair amount of database caching.  Just
intuitively, having 17 disk arms servicing thousands of interactive
users strikes me as unreasonable.  Even if you can cache many of the
reads, there should be enough writes to peg the arms.

Sometimes management has difficulty understanding the disk arm thing
because it seems to be not as critical on other platforms.

Regards,
Andy

> On Behalf Of Charly Jones
> Subject: Re: Disk arms and faulting: (was - We've Added more
memory...but
> I can't remember!)
>
>
> Andy --
>
> I am going to respond to your questions in installments, because I am
> multi-tasking, and because the questions are complex.