 |
 |
Archives of the TeradataForum
Message Posted: Sun, 10 Nov 2002 @ 22:50:48 GMT
| Subj: | | Re: 3NF vs Dimensional modelling |
| |
| From: | | Neil Raden |
You wrote:
| | 3NF is preferable whenever it can deliver the required performance, since it's far more flexible. Teradata is unique in that it
often can deliver acceptable performance with 3NF data bases. | |
Pat, can you be precise and rigorous when you say 3NF is "far more flexible"? Can you prove it? I don't think you can, it's just
an idea that no one has ever proven. Secondly, a dimensional schema, with normalized dimension tables (also know as a snowflake schema) is,
in fact, in 3NF. Does that mean you endorse a dimensional schema?
My point is, there are lots of designs that are 3NF, and most of them will not perform at all. Which ones will? And just what is it
that makes a 3NF so flexible? Please explain WHAT KIND of 3NF schema you are recommending and why various design choices lead to this
flexibility.
-NR
Neil Raden
Hired Brains, Inc.
www.hiredbrains.com
| |
