For a sort to make predictive assignments of resources you would need a
function like Global DSM such as Syncsort has.  They have had that
capability for over 15 years.  It considers what is going on right now
and what has gone on in the past during the time it expects the sort be
running.  This makes it so that if you start a sort at 08:55 and
everyone signs onto TSO at 09:00, if the sort is going to run over 5
minutes it will get fewer resources than if it was going to finish
before 09:00.


Christopher Y. Blaicher
BMC Software, Inc.
Austin Development Labs
(512) 340-6154
BMC Software, Inc. makes no representations or promises regarding the
reliability, completeness, or accuracy of the information provided in
this discussion; all readers agree not to rely on this information or
take any action against BMC Software in response to this information.


-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On
Behalf Of Hal Merritt
Sent: Tuesday, January 17, 2006 12:07 PM
To: [email protected]
Subject: Re: DFSORT performance data

Point is that it does. More, it makes its decisions pragmatically based
on the reality at that point in time.  

No such process can get it 100% right %100 of the time. But it should
consistently do it far better than a human after the fact. For one
thing, it is difficult for a human to weigh all of the factors in a
constantly shifting environment.  

Consider: SORT A kicks off, talks a look at the environment, and elects
to use x resources. Process B kicks off and drives up the system paging
rate. Sort C now kicks off, sees the problem, and elects to use a
different mix of resources. Again, SORT C is looking at the environment
and the nature of the task at hand and will try to make the most
efficient use of the resources available. Enter Process D which again
drives up the system page rate. Now SORT E has a completely new
challenge. 

After the fact, we might second guess what SORT A did. But is that
really what we want to do? Or should we consider the collective behavior
of all of the sorts in the mix?

Just my thoughts. 

HTH and good luck. 

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to