ADSM-L

Needs assessment.

2000-02-25 15:24:34
Subject: Needs assessment.
From: "Allen S. Rout" <asr AT NERSP.NERDC.UFL DOT EDU>
Date: Fri, 25 Feb 2000 15:24:34 -0500
Greetings all.


I'm working on a vaguely rigorous answer to the question "how much of the
capacity of my ADSM (smack) TSM infrastructure am I really using?".

Much of this is straightforward; counting tapes, counting time, adding
incremental sizes and dividing by disk...

But I'm having a lot of trouble really _measuring_ how many "tape-hours" I'm
using every day.  Is anyone else doing this?

This is my thinking:

     Activity:                                  Cost (tape-hours)

Backups that run direectly to tape              (session_duration) * 1
(yes, I still have a few of them.)

Database backup:                                (finish - start) * 1

Backing up a disk pool:                         (finish - start) * 1
Migrating disk to tape:                         (finish - start) * 1

Backing up a tape pool:                         (finish - start) * 2
Reclaiming a tape pool:                         (finish - start) * 2


Do you see the problem?  (*spoiler below*)


Depending on when different processes fire, I could be reclaiming or migrating
several different pools (or trying to) at the same time.  It so happens that I
have two 3590 drives (3494s rock!)  and I regularly see reclamation processes
three deep.

A cursory examination of the activity logs does not appear to provide any
additional granularity; "time I actually _got_ my mount points" doesn't seem
to be in there.

I am trying to avoid doing a 1-minute poll of Q MOUNT and taking care of it
that way.  (not the least because I can then only _begin_ to collect data now,
and I'd really like to derive my answers from extant logs.)


I've looked for answers to this, I'll graciously accept clues along the lines
of "Dumpkoff, look at page 34-B of the redbook on analyzing your ADSM usage"
or such. :)


Allen S. Rout
<Prev in Thread] Current Thread [Next in Thread>
  • Needs assessment., Allen S. Rout <=