HP3000-L Archives

April 2005, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Mark Landin <[log in to unmask]>
Reply To:
Mark Landin <[log in to unmask]>
Date:
Tue, 12 Apr 2005 09:26:58 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (36 lines)
Elongation. An elongation of 1 is optimal. Any elongation above 1
indicates that you are, on the average, doing more I/O than is
necessary when you go traversing that particular chain, because the
records that comprise that particular chain are not packed together in
the same blocks.

When you do a DETPACK, which puts all the records on a particular
chain into as few blocks as possible (thus minimizing I/O) you have to
pick which path to pack along. Pick the one you want to minimize I/O's
on, which will typically, but not always, be your primary path.
Optimizing along one path will often inflate elongation in the other
paths, so choose wisely.


On Apr 12, 2005 9:17 AM, Jay Willis <[log in to unmask]> wrote:
> ...that would indicate that a dataset needs to be re-packed?
>
> I would have sent this question to Robelle, except that we dropped support.
> Budget cuts. I know that doing a detpack will reduce the highwater mark, but
> are there any other stats from the report that indicate that performance
> would in increased with a detpack? We (3-man, 1-woman shop) want to come up
> with a job that possibly would run every night to run Howmessy for important
> databases and send an alert (email) if such-and-such a stat was getting away
> from optimal and that a detpack or capacity change needed to be done. I
> understand when a dataset's capacity needs to be increased, but not when a
> detpack needs to be done (except when the highwater mark of a detail would
> suggest it). We'd rather not do a detpack on a regular basis - just when one
> is indicated.
>
> * To join/leave the list, search archives, change list settings, *
> * etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
>

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

ATOM RSS1 RSS2