Distributing the load across SORTWORK files

[Guy Werry]

Hey!  I agree too, Ken.  I use Qtp for almost all of my data extract stuff because it is SO much more powerful and convenient to create a subfile in Qtp than Quiz.  Especially true if you have to do any sort of data manipulation / totalling, etc.
 
Guy L. Werry
Senior Systems Analyst
Hudson Bay Mining & Smelting Co., Limited. 

-----Original Message-----
From: Ken Langendock [mailto:Ken at Langendock.com]
Sent: Wednesday, May 17, 2006 3:27 PM
To: Guy Werry; powerh-l at lists.sowder.com
Subject: RE: Distributing the load across SORTWORK files


I agree with Guy
 
When I wrote the sales system for Carling O'Keefe, we were sorting millions of records and we reduced the report job from 24 hours to under 20 minutes by using the SORT command at the OS level on a Vax.
To sort the subfile at the OS level we had to exit QUIZ or QTP between executes, but it was well worth the effort.
 
On the other hand, I still say extract all your data in QTP using Cursors and output phase reads and use QUIZ as a DUMB subfile reporting tool will definitely improve the performance of an application and still be portable from one platform to another..
 
A combination of the two would be the ultimate approach.
 
Ken
 
-----Original Message-----
From: powerh-l-bounces+ken.langendock=rogers.com at lists.sowder.com [mailto:powerh-l-bounces+ken.langendock=rogers.com at lists.sowder.com] On Behalf Of Guy Werry
Sent: May 17, 2006 3:35 PM
To: powerh-l at lists.sowder.com
Subject: RE: Distributing the load across SORTWORK files


John,
 
My world is Hp-Ux and before that MPE-ix, but I think the advice would still apply.
 
The ideal way to do something like this is to write to and INDEXED SUBFILE, which has the index in the order that you require.
 
That way the records are "inserted" (written) in the order of the index.  A quiz report can then just read the resulting subfile without having to do the sort at all.  Yes, the process that creates the subfile is slow, but the report is very fast and the overall load on the system is less.
 
Failing that, if you don't have indexed subfiles available I would seriously consider putting another file definition into the dictionary.
 
Other than that, a system sort utility would be a good idea.
 
Guy L. Werry
Senior Systems Analyst
Hudson Bay Mining & Smelting Co., Limited. 

-----Original Message-----
From: powerh-l-bounces+guy.werry=hbms.ca at lists.sowder.com [mailto:powerh-l-bounces+guy.werry=hbms.ca at lists.sowder.com]On Behalf Of John Stires
Sent: Wednesday, May 17, 2006 2:01 PM
To: powerh-l at lists.sowder.com
Subject: Distributing the load across SORTWORK files


We are doing a very large sort in quiz.  It consistently runs for a time and then dies.  We have 10 SORTWORK files defined, but only one seems to be used.  It grows to several million blocks before the process dies.  The others are virtually untouched.
 
We are running OpenVMS V6.2 with PowerHouse version 7.10E6.
 
Does anyone have a suggestion on how to distribute the sort load across multiple SORTWORK files.
 
 
Thanks in advance for any help any of you can offer on this.
 
John Stires

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sowder.com/pipermail/powerh-l/attachments/20060517/6e9f1da2/attachment.htm