Re: Rookie asks about "Normalization"

From: Renato Brazioli <brazioli_at_sodalia.it>
Date: 1995/12/06
Message-ID: <4a46dp$7v3_at_server-b.cs.interbusiness.it>#1/1


roywagner_at_aol.com (Roy Wagner) wrote:
>>a completely normalized app.would be Ideal but is not always the best
>>
> This is an absolutely true statement.
>
>In very simiple terms, a normalized database contains no redundant
>information. Any repeated information is pointed to using, usually an
>number that is the primary key of the table containing that type of
>information (once and only once).

This is just partially true. It is not always a number, more often is a composite key. Adding a numeric, unique identifier (AKA Surrogate Key) is already a denormalization.  

 It is indeed true that normalization is very good in an ideal world but real world problems could cause you to move away from it. It is important that this happens in a controlled way, i.e. knowing the denormalization that is applied. It is also true that modern databases mechanisms, like triggers, help a lot to enforce data integrity even whendenormalization is present.

 Just my opinion ...

  Renato Brazioli

-- 

Renato Brazioli      Sodalia SpA              e-mail: brazioli_at_sodalia.it
                     Via del Brennero, 364    voice:  +39(461)316.154 (direct)
                     I-38100 Trento           fax:    +39(461)316.127
                     ITALY
Received on Wed Dec 06 1995 - 00:00:00 CET

Original text of this message