Re: Update file definition with a LOT of records - best approach?

["Booth Martin" <Booth@xxxxxxxxxxxx>]

 You are doing all of the records?  100% of them?  Why are you processing by
key? Take the K off the F-spec.


---------------------------------------------------------
Booth Martin   http://www.MartinVT.com
Booth@MartinVT.com
---------------------------------------------------------

-------Original Message-------

From: midrange-l@midrange.com
Date: Monday, November 11, 2002 02:37:49 PM
To: midrange-l@midrange.com
Subject: Update file definition with a LOT of records - best approach?

--
[ Picked text/plain from multipart/alternative ]
Hello folks,

Our product has a main file (with a unique key on the PF, plus a LOT of
logical files) that can grow to pretty large record volume (25M to 200M
records). We are needing to change the way we do a new release with a
database change to that file. The question I have is this - what is the best
way to update the database file definition and translate the data to the new
definition for a large record count?

CHGPF changes the file "serially" which is terribly slow. I have written a
process that submits a given number of "parallel" CPYF commands, but it
appears that the constraint is in building the UNIQUE access path of the PF.
Once I get to 5 or 6 jobs for a 3.5M record test data set on our
two-processor development box, I cease to get runtime improvement. It takes
1:20 plus/minus 3 minutes. (an hour and 20 minutes) I have removed all
logical files so that the only index being built is the UNIQUE one on the
physical file.

I hope the question is clear. What do you folks think?

Thanks in advance,
Michael Polutta
Atlanta, GA
_______________________________________________
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@midrange.com
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/cgi-bin/listinfo/midrange-l
or email: MIDRANGE-L-request@midrange.com
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.

.