Re: Maximum of Columns in one table (9i)

tina london wrote:
> "DA Morgan" <damorgan_at_psoug.org> wrote in message
> > Per sane design 40-50 column max. Above that number it is time to
> > get someone to critique your design.
>
> surely 40-50 columns is way too high and indicates a denormalised design,
> not that necessarily there is anything wrong with that, I would like to
> suggest anything with more that 10-14 columns is probably in need of
> attention.

This is getting silly. The right number of columns for a table is not "under 14" or "under 50", it is whatever number of columns you get after normalisation. If every column is functionally dependent on the key then the table is normalised, even if it does contain 17 columns or 57.
>
> step 1 produce a normalised model
>
> step 2 denormalise to get the performance needed in critical areas

Denormalisation for performance is SO overrated. I'm not saying never, EVER do it, but I will say that in my experience 99 times out of 100 when I see cases of "denormalisation for performance" it is a poor decision that doesn't actually achieve worthwhile performance gain. Also, I have NEVER yet found a situation where I have needed to do it. Received on Wed Jul 06 2005 - 09:01:07 CDT