Re: Newbie question about db normalization theory: redundant keys OK?

David Cressey wrote:
> "Dr. Dweeb" <spam_at_dweeb.net> wrote in message
> news:476593b3$0$21933$157c6196_at_dreader1.cybercity.dk...
>

>>>Would you a) tell the university that the data cannot be modelled and
>>>you can't hold it in a database or b) use an artificial key?
>>
>>This makes my student number 7613861 artifical?
>>This makes my employee number 644923 artificial?
>>
>>One is used by my university, the other by my employer. They may be
>>arbitrarily assigned, but they are unique and indicate me alone.

As was your name when (presumably) your parents assigned it to you.

> The assignment of an as yet previously unused number to you might have been
> done by a clerk in the registrar's office or human resources department.
> More likely, it was done by application software lying somewhere in the
> processing chain between data entry and insertion into the database.
>
> It's when the application software relies on the DBMS to come up with a
> previously unused number that things start to get muddy. As long as the
> assignment has been done before the DBMS ever sees the data, the
> distinction between "natural" and "artificial" is moot, and the database
> design might as well consider those two numbers to be "natural keys" of you,
> in two different roles.
>
> "Artificial" and "surrogate" are often used interchangeably. They shouldn't
> be.
Received on Mon Dec 17 2007 - 16:31:46 CET