Re: which data model is better?

Daniel Morgan wrote:

> I think one can safely say, as a generalization, that flat file models
> are faster than relational models. The "compelling" reason I refer to

As a generalization, I would say "flat file models are faster than relational models ... for data retrieval". I find the "denormalization for performance" discussion generally falls flat when looking at the entire application, including DML, across the entire user population. Hence the need for snapshots, MVs, or equivalent.

(Then again, I am of the opinion that each and every index should be reviewed every 6 months to verify that it is still required. After all, how often has an index been created to speed up a report, only to have the report march off into the sunset - never to be used again. And yes, some will always have a check beside the "keep" box.)

In any case, the answer is 'depends - with better details provided during realistic benchmark/simulation of this environment.'

/Hans Received on Sat Jan 17 2004 - 22:22:44 CST