Re: A database design question

That's good advice.
Most systems I've seen tend to stop at 3rd normal form, but remember at that level multiple schemas, that contain the same design information are possible. So there is no one right way at 3rd normal form ... There are a few systems that have "stated" they were at 4th or 5th ... but it is a horrendous amount of work to get some basic information out. One designer on the project felt it was like building a car from a box of nuts and bolts everytime ... even though all they ever dealt with was cars. One person that I learned allot of excellent design guidelines from would always come back to some basic principles ... does the design easily represent the intent of each entity. I find that useful to remember as well...
Thor HW

terryg8 <trg_at_ibm.net> wrote in article <33E14B34.4516_at_ibm.net>...
> On Wed, 30 Jul 1997 15:07:04 -0500, Sanjay Rallapally
> <c696346_at_showme.missouri.edu> wrote:
>
> >
> >Its me again,
> >I am curious abut the level of database&nbsp; normalization that is
> >followd in the real-world . How rigidly do u guys follow the database
 design
> >rules regarding normalization? Lastly, when is it better to have an
 unnormalized
> >table, than an unnormalized one? I mean, how useful are indexes as
 against
> >having a normalizzed table
> >
>
> In my experience, the issue of proper normalization sometimes
> hides faulty designs. You can have real garbage interface
> driven database designs that are technically perfect 3rd normal
> form. That buys you very little if the development effort
> hasen't paid due regard to the fundamental concepts and
> physical realities of the systems being modelled.
>
> My opinion is that normalization isn't an issue of right
> or wrong. System constraints will always dictate what
> is appropriate. Worry foremost that the system model
> will implement lasting entities and processes that won't
> change with the whims of the next developer that comes along.
>
> My 2 cents worth. In any case, good luck.
>
> TRG
>
>
>
Received on Fri Aug 01 1997 - 00:00:00 CDT