Re: Data Model

From: David Cressey <dcressey_at_verizon.net>
Date: Sat, 04 Mar 2006 13:22:14 GMT
Message-ID: <awgOf.442$6h1.333_at_trndny09>


<matthewdavis1980_at_hotmail.com> wrote in message news:1141443096.151369.230200_at_z34g2000cwc.googlegroups.com...
> Frank,
>
> I'm curious as to why you can appreciate that a single table was used.
> Not that I specifically agree or disagree, I'm open, I would just like
> to know your reasoning behind it.
>

PMFJI. For years, I have been really enthusiastic about Ralph Kimball's Star Schema design for building a multidimensional model projected onto the SQL-RDM. I would still recommend it as a possible design choice for anyone building a Db where OLAP takes precedence over OLTP. For OLTP, I'd prefer a more normalized design.

If you take a close look at a Star Schema, it's really only a few steps removed from the "one big table" concept. It removes all of the dimensional attributes from fact tables, and replaces them with a foreign key to a dimension table. But it's really almost one big table.

So much so, that it's easy to build a view of a star schema that shows it as one big table, and then do all your queries as subsets of that one big view. Not necessarily the best implementation, but great for prototyping. Received on Sat Mar 04 2006 - 14:22:14 CET

Original text of this message