Re: Database schema for univesal usage

Jan Hidders wrote:
> lauri.pietarinen_at_atbusiness.com wrote:
> >
> > I actually agree with your viewpoint and I personally tend to design
> > databases that are tightly fit.
>
> Recently I came into contact with a colleague of my university who
> actually does research on the differences between more and less-generic
> data models. One of his experiments was designed to test the hypothesis
> that more generic data models would be easier to maintain. To be more
> specific, he looked at the case where the data model has to be extended
> because of new requirements. He found that for more generic data models
> this sometimes actually becomes harder, i.e., it took more time for the
> designer to understand the old model and come up with the new design,
> and they made more mistakes. That is a cost that is often overlooked.

Very interesting. Could you give the name of your colleague so that I can find it when it comes available?

The reason for using generic (or less spesific) schemas (at least one of them) is to avoid the need to change the schema in the first place. The $$$ that need to be spent on adding a new column is not because of the ALTER TABLE -statement, of course, but because of the cost of the plumbing needed to get the data of the new column to the user interface.

Additionally, big enterprises typically have several separate systems between which they pump data. The pumping code and interface (XML or not) has to be modified and tested etc... If no modifications to the schema are involved this does not need to be done.

Regards,
Lauri Pietarinen Received on Sun Jun 05 2005 - 07:35:32 CEST