 |
 |
Archives of the TeradataForum
Message Posted: Wed, 13 Nov 2002 @ 11:28:22 GMT
| Subj: | | Re: 3NF vs Dimensional modelling |
| |
| From: | | Ken Hansen |
Duncan, thanks for your reminder - I was indeed thinking of 3NF EDW as compared to cubes in multidimensional DBMS's. A
Dimensional model in a relational database is far better and may suffice if, as you say, "the enterprise is not complex".
Neil, apologies for thinking of cubes but, perhaps because I am primarily a data user rather than a modeller, my experience runs
counter to assertions that Dimensional Models reflect the way business people think - the very reason that I advocate 3NF EDW is
that, although Dimensional Models may reflect the way business people think of particular report(s) at one moment of time, that
thinking will evolve rapidly and often. That is why the enterprise is best served by holding core data at atomic level in the core
DW - sooner or later someone is likely to want/need that analysed and/or summarised by changed or new dimensions that may not yet be
identifiable.
Ken Hansen
| |