HP3000-L Archives

August 2000, Week 5

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:
Chris Goodey <[log in to unmask]>
Reply To:
Chris Goodey <[log in to unmask]>
Date:
Wed, 30 Aug 2000 16:52:50 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (76 lines)
Poor end of month performance?

It would be nice to have a few details.

Often I see shops trying to run way too many jobs. If I toss on
8 more killer jobs at end of month, things are going to slow down!

They may be out of memory, and adding an extra job or two pushes
them over the edge, and performance suffers. Some cheap third
party memory would help. If they have Glance, use it, if not, buy it!
This can help you figure out what is really going on.

CPU upgrades can be real expensive if you have products like Cognos.
I just upgraded from a 977SX to a 979-200 and most all of the
software made the upgrade with no additional charges!
However, some products can really cost you!

It may be that a simple data base reorg would help lots.
I had one application that slowed down as more records were added
to already long chains. The program spent lots of time searching
all over the disc for the various records. A reorg that took a few
hours made the application run several times faster! So we would
reorg it on a regular basis.

I have seen other data bases with lots of real old data that wasn't needed.
Simply deleting some old data and shrinking the data base helped immensely.

Other places had one single job that was the main performance killer.
Re-writing the job more efficiently solved the problem.

If you have a small amount of memory, like 256mb or less,
it is pretty cheap to try adding some more. However, a 967 has a performance
rating of 2.6, compared to the HP suggested upgrade of a 979 at 7.9,
so it may be you are ready for a faster system. If Glance shows you
as mostly CPU limited, this may be your only choice. On a small memory
system a fair amount of cpu is used just swapping data, so more memory can
actually give you more cpu.

Look at the jobs and see what kind of resources they are taking. If you have
jobs
using cpu hours, then they need to be re-written (not always feasable),
rescheduled
for a later time, or you need a faster system. If you are trying to run more
than 1 or 2 heavy jobs, then just limiting it to one at a time might
increase your total through-put as 2 jobs no longer fight each other
for resources.

This free advice worth what you paid for it :-)

-----Original Message-----
From: Paul H Christidis [mailto:[log in to unmask]]
Sent: Wednesday, August 30, 2000 4:20 PM
To: [log in to unmask]
Subject: [HP3000-L] Multiple HP3K related questions


Fellow list members,

My latest tasks have brought forth a few questions that I'd like to bounce
against the collective knowledge base:

1

3.  On occasion I perform some remote support on an HP967SX located in
another state.  The users have been complaining, especially around
'financial month end' about the poor response (in terms of length of time
for jobs to be processed), and their management are willing to examine the
available options.

Keeping in mind that they'd like to stay within the same tier, in order to
avoid the big upgrade cost of a certain vendor, can anyone offers some
suggestions.

Regards
Paul Christidis

ATOM RSS1 RSS2