Re: DB Design Principles for RDBMS

From: Richard Wheeldon <richard_at_rswheeldon.com>
Date: Fri, 14 Mar 2003 00:22:01 +0000
Message-ID: <3E7120A9.59E2_at_rswheeldon.com>


RayD wrote:
> Not to mention that a lot of people misuse the term denormalize.
> Storing derived or aggregate data is not denormilization. To the
> contrary, the tables derived data is stored in are usually perfectly
> normalized. They just contain derived data, like an account balance or
> something.

Help! I'm confused. How can you have a 5NF schema with derived data? I thought 5NF required no further possible decomposition without loss of data. If the data can be derived, it is  (a) possible for this data to be removed without loss.  (b) possible for this data to be inconsistent. So how can this be fully normalized ?

Is normalization ever affected by whether the function to derive the data is something that the database can do, or something which requires external knowledge ?

Richard Received on Fri Mar 14 2003 - 01:22:01 CET

Original text of this message