Re: Len Silverston's Universal Data Models sanity

On Tue, 04 May 2004 23:46:30 -0300,
=?iso-8859-1?q?Leandro_Guimar=E3es_Faria_Corsetti_Dutra?= <leandro_at_dutra.fastmail.fm> wrote:

>> and they are highly redundant.
>
> Do you mean, not normalised?

No, I mean that they have a lot of data that could be derived from other data like article stocks.

>> For instance the stock of an article could be derived from the
>> inputs and the outputs of the article.
>
> Yes, but there are performance considerations there. For
>example, how do you keep the system from degrading after some years of
>highly movemented goods?

Physical data independence is for that. The DBMS must ensure that the sum of the movements matches with the stocks. If not, you would have uncontrolled redundancy. To control this redundancy has the same performance cost as to not have logical redundancy at all.

I most systems I know, any application might corrupt the database and sometimes they do!

It is very difficult to guarantee the integrity of the data if the data derivation is implemented using thousands of triggered procedures. This is probably my primary motivation for the development of a TRDBMS. I have hundreds of views derived from a few movements base tables.

You can derive all the information about the state of your business from a few base tables. It would be great if we could declare all the inference rules without losing performance.

>> I have several designs I could post, but I have to translate them
>> first.
>
> No need to translate, Brazilians can understand Spanish quite
>well... it would be nice to set somethink like a wiki for that.

But there are many readers in the group who don't speak latin languages ;-)

>Problem would become maintaining it, and settling on a preferred
>format. I would say Tutorial D would be quite nice for that, D4
>good enough

I agree, but I am more comfortable with Tutorial D.

BTW What do you think about to use Tutorial D or D4 instead of XML for data exchange?

Regards
  Alfredo Received on Wed May 05 2004 - 16:31:30 CEST