Re: Normalization ?

Hi David,

I agree with you entirely. When I mentioned denormalize I sort of meant denormalizing
it to make it optimal to use for the purpose intended. And you have in fact hit the nail
on the head when you mention dimensional modeling because that is exactly the kind
of purpose we have for this second read-only database.

Dimensional modeling is not my area of expertise so I am greatful to you for the reference
to Kimbal's book. I will certainly follow this up.

So in principal you do not see any particular problems/issue in arranging these two databases
as outlined in my initial scenario, i.e.

    Database A an OLTP, normalized DB feeding data into     Database B organized/optimized for use as a read-only repository of 'good' data from A?

Thanks for your reply,
Naran.

David Cressey wrote:

> > I'd be very interested to hear what other perspectives people have in
> > this NG.
> > If possible, please CC your replies to me as well as posting to this NG.
>
> Naran,
>
> I wouldn't denormalize "just to denormalize". What I would do is use some
> design principle to help me put things that
> might go well in the same table together. If that design principle
> contradicted what normalization would dictate, so be it.
>
> A specific design principle that has helped me in a lot of these situations
> is Dimensional Modeling: combining data into fact tables and dimension
> tables, arranged in a star schema. If all this makes no sense to you, you
> can check out one of Kimball's books on Data Warehousing. see
> www.rkimball.com
>
> --
> Regards,
> David Cressey
> www.dcressey.com
Received on Tue Sep 11 2001 - 10:01:50 CEST