Re: Feedback on approach requested

["Dave Odom" <Dave.Odom@xxxxxxxxxxxx>]

A word to the wise... recreate the iSeries flat file to a 3rd normal
form set of tables and you can drastically reduce the update problems
found with flat files.   

The trigger idea is a sound one providing there are not TOO many
"triggerings" updating the second set of tables(which is what I'd
recommend vs. flat files for lots of queries) on a production(assumption
on my part) system.  If the updates to the summary files/tables start to
cause performance problems, it may be you are at a point wherein you
need to separate your query summary tables from your production
transaction files/table and turn the summary, I assume used mostly for
queries/decision support/what if situations, into a data mart or data
warehouse on another LPAR or box where queries/decision support/what ifs
are a speciality.   The update loading of the data mart/data warehouse
is done differently than a transaction environment.   But, I'm sure you
already know this.    

Just some thoughts.

Take care,

Dave Odom
Arizona       

>>> rob@xxxxxxxxx 10/11/2005 11:55:16 >>>
Suppose you had to query this file that had 2 million records.  In it
is a 
load number and a date, perhaps state, etc.  And every time you 
added/updated/deleted a record in this database, it would automatically

update another file that was summarized.  Perhaps only three fields:
Date, 
number of loads, state.  Or some such combination.  You could add a 
trigger to the 2mil record file that would automatically update this 
summary file.  That way you would not have to modify one line of any of

the program(s) that update the big file.

Rob Berendt