Re: Normalising a two-to-one relationship

Philip Lijnzaad wrote in message ...
>
>
>David> Remember, normalization is a design principle, not an absolute
 golden
>David> rule.
>
>Brief and perhaps off topic note: some hold that normalization is part of
 the
>_analysis_, not the _design_ . I.e., normalization is something that is
>inherent in the part of real world (or 'business logic') that is being
>modeled. The analyst's task is to uncover this web of idealized
>relationships, then map it to some (usually less idealized) form.
>
>
Philip

I agree, normalization should be part of the analysis. However, the need for denormalizing tend to show up in the design phase. Hence, I find it a sound principle that one should complete a normalized model before considering any denormalizations.

--
Roy Brokvam