Re: Normalization by Composing, not just Decomposing

From: Alan <>
Date: Tue, 13 Apr 2004 12:32:56 -0400
Message-ID: <c5h4n1$1if1u$>

Powerplay is optimized to build a cube from a flat file (which would have been extracted from a star schema). I view the start schema as a variation of the relational model. The constructs and constraints are the same, their just assembled differently. It's a relational model that follows rules other than normalization rules, IMO...

"Laconic2" <> wrote in message
> Alan,
> I once built a star schema to validate a powerplay cube
> (by providing an alternate route for the data from the OLTP system into
> cube.)
> Here's what I didn't expect: it took less time top copy the data from the
> OLTP system to the star schema, and then from the star schema into the
> than it took to load the cube directly from the OLTP data. I don't know
> whether that was just a fluke or was to be expected.
> Interesting discussion you are having concerning dimensional modeling,
> relational modeling and star schemas.
> I think of a star schema as a projection of a dimensional model into the
> world of tables, columns etc. I don't think of a star schema as having
> particularly derived from a relational model.
Received on Tue Apr 13 2004 - 18:32:56 CEST

Original text of this message