|
|
Archives of the TeradataForum
Message Posted: Mon, 11 Nov 2002 @ 19:04:27 GMT
Subj: | | Re: 3NF vs Dimensional modelling |
|
From: | | Belle, Patrick A |
Neil,
The flexibility of 3nf is inherent in its mathematical underpinnings. In the original papers issued over 30 years ago, the mission of
the entire design was to make all data requests possible with a minimum of update overhead. When you relax the normalizations, you
introduce update redundancies which are often overlooked in the ETL resulting in anomalies in the data base. If they are not overlooked,
the ETL Process is often burdened by the effort associated with maintaining these redundancies.
As to your assertion that a dimensional schema with normalized dimension tables is in fact 3nf, I say "Can you prove it?". I don't
believe it for a minute.
And in terms of what design I endorse, I endorse what works in the situation at hand and this often depends on the resources (hardware,
software and personnel) available as well as the overall objectives of the engagement. I don't have a litmus test, but I'll favor 3nf every
time all else being equal.
Pat Belle
Teradata Certified Master
| |