Excuse the formatting, I hacked this out of a digest........

My thought is that the hashing algorithm is causing a clustering situation
in your file. The filename indicates to me that this is some kind of check
file. Is your key a check number and/or a check date? If you write a lot of
checks on a certain day of the month and not others, this could be is
what is causing the problem.

HTH,

Pat Shugart
Oaksoft Consulting, Inc.
(248) 489-6565

========================================

Date:    Mon, 1 Dec 1997 13:32:56 +0000
From:    Bryan O'Halloran <[log in to unmask]>
Subject: Re: NM KSAM question
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

     Lars and Costas advised some very good rules to shorten NM KSAM
     load jobs.  I am following most of their rules except that I use
     SUPRTOOL rather than FCOPY to load the file.

     One of our load jobs always takes a long time with the last 75%
     of the load taking much longer than the first 25%.  An example is
     set out below

     SUPRTOOL/iX/Copyright Robelle Consulting Ltd. 1981-1997.  Type H
     for help.
     (Version 4.0.10 Pre-Release)  SAT, NOV 29, 1997,  4:31 AM
     Licensee:  Mercantile Mutual Life Insurance [114G]
     >>>> INPUT CHEQFILE
     > OUTPUT *NEWCHEQ
     > EXIT
     Starting combined input and output phase.
      25% In=141642     Out=141642    Delta-Min=3:43. Wall-Min=3:43.
     Cpu-Sec=176.
      50% In=285846     Out=285846    Delta-Min=45:44 Wall-Min=49:27.
     Cpu-Sec=2497.
      75% In=430050     Out=430050    Delta-Min=49:17 Wall-Min=98:44.
     Cpu-Sec=5049.
     100% In=577090     Out=577090    Delta-Min=41:29 Wall-Min=140:13.
     Cpu-Sec=6916.
     IN=577090, OUT=577090. CPU-Sec=6919. Wall-Sec=8592.
     Wall-Min=143:12.


     END OF PROGRAM

     Can someone explain why it takes over 40 minutes to load the
     second (and subsequent) quarter(s) of the input rather than the
     less than 4 minutes to load the first quarter.

     Thanks

     Bryan O'Halloran