Re: Normalization

From: Leandro Guimarães Faria Corsetti Dutra <leandrod_at_mac.com>
Date: Tue, 02 Oct 2001 12:55:13 -0300
Message-ID: <3BB9E361.4070702_at_mac.com>


Bjørn Kr. Sveinsvoll wrote:

> Hi, everybody!
>
> I am looking for a web-site, document, book that can guide me in the
> normalization process. I have several tables but when it comes to
> whether or not I should normalize them I don't know what to do. I feel
> a bit Codd-dyslexic. I know how to do it, but I don't. :-)

        http://dbdebunk.com./ has pointers to articles and books.

        Please note that there is a great confusion over this. The relational theory -- from whence comes the idea of normalization -- is a logical one, that has nothing to do with physical implementation (how the bits are organized on disk and memory, and how to map logical constructs to physical ones).

        Therefore, you can normalize at the logical level, but what's called denormalization is to be done at the physical level -- so it shouldn't be called denormalization, but physical design.

        It seems that this confusion arose because SQL -- which isn't really relational! -- doesn't support the separation of logical and physical levels as it should.

        So the answer is, you should always normalize, and test as it is. If needs be, you keep your logical normalized docs and concepts, and change the physical design -- just don't call it denormalization.

-- 
  _
/ \ Leandro Guimarães Faria Corsetti Dutra           +55 (11) 246 96 07
\ / http://homepage.mac.com./leandrod/     BRASIL    +55 (43) 322 89 71
  X  http://tutoriald.sourceforge.net./     mailto:lgcdutra_at_terra.com.br
/ \ Campanha fita ASCII, contra correio HTML    mailto:leandrod_at_mac.com
Received on Tue Oct 02 2001 - 17:55:13 CEST

Original text of this message