Re: database design method

From: Leandro Guimarães Faria Corsetti Dutra <lgcdutra_at_terra.com.br>
Date: Fri, 25 Oct 2002 12:45:33 +0200
Message-ID: <apb7ce$obu$1_at_ID-148886.news.dfncis.de>


David Cressey wrote:

>> It seems to me that most theoretical texts talk about 'conceptual',
>>  'logical' and 'physical' design. These steps produce conceptual,
>> logical

>
> and
>
>> database schema. However in the 'real world' conceptual and logical
>> steps seem to be combined into a single step.

>
> Here's my take on it. A "conceptual model" is typically the result
> of analysis, not design. The best conceptual models capture the
> information requirements of some proposed system or database, from a
> data centric point of view. The ER model, when it was first proposed
> and for about a decade following, was a good model for this purpose,
> because of the way relationships are treated. In the ER model,
> relationships are identified, but not implemented.

        Wrong, ER can't capture all of the organisation's rules, because it has no equivalent to all integrity constraints in the database.

        Actually, conceptual model would be the rules derived from analysis. A logical model would be the realisation of that as a relational database.

> A Relational DBMS may have pointers and keys associated together in
> indexes, but these are managed by the RDBMS itself, and are nearly
> transparent to the application programmer.

        Should be not "nearly", but "totally" transparent.

> So what's the difference between a conceptual (ER) model, and a
> logical (Relational) model? Well, one difference is how tied to a
> relational implementation you are. A relational model is somewhat
> more bound to a relational physical implementation than an ER model
> is.

        Wrong again, the physical model is determined by the DBA from both the logical model and the system implementation physical constraints and utilisation requirements. On the other direction, the logical model is totally independent from the physical implementation.

> But the big difference, in my mind, is the difference between
> analysis and design. The conceptual model captures the analysis,
> while the logical model captures the logical design. I expect to hear
> a lot of contending views on this topic.

        Actually this is what you got kinda right. Received on Fri Oct 25 2002 - 12:45:33 CEST

Original text of this message