Re: Specifying all biz rules in relational data

From: Kenneth Downs <firstinit.lastname_at_lastnameplusfam.net>
Date: Tue, 21 Sep 2004 22:24:34 -0400
Message-ID: <2pnqic.44v.ln_at_mercury.downsfam.net>


mAsterdam wrote:

> robert wrote:

>> ...  the batch processing
>> COBOL folk will simply not allow even 3rd normal form structures.  they
>> simply don't chew as fast as VSAM flat files for that kind of processing.

>
> What excuse do they have for batch processing
> for anything beyond garbage collecting, reporting
> and [end|start]-of-[day|week|month|year]
> aggregation?

Marketing databases benefit greatly from batch processing, as I would imagine would any largely static database.

Every three months you do an NCOA on a marketing database, National Change of Address. Reading and writing 10 million rows in a transactional database is very bad news. You just don't need ACIDity on a row-by-row basis. What we do is dump the file to disk, run it through the process, erase the tables, and reload them. This is three complete read/write passes and it is still 100 times faster than buying the overhead of ACIDity on the updates.

-- 
Kenneth Downs
Use first initial plus last name at last name plus literal "fam.net" to
email me
Received on Wed Sep 22 2004 - 04:24:34 CEST

Original text of this message