Re: Normalization, Natural Keys, Surrogate Keys
Date: Thu, 23 May 2002 18:57:46 +0100
True, and It's also worth bearing in mind that "natural" keys are more likely to have natural partition lines that surrogate keys.
How many times have we seen the complaint "I want to partition this data by date, but I have a surrogate key and Oracle only allows me to have a local primary key index if it contains the partitioning column."
Every time I've seen the problem, the obvious partitioning element has been part of the natural key - which has been replaced with a meaningless number.
-- Jonathan Lewis http://www.jlcomp.demon.co.uk Author of: Practical Oracle 8i: Building Efficient Databases Next Seminar - Australia - July/August http://www.jlcomp.demon.co.uk/seminar.html Host to The Co-Operative Oracle Users' FAQ http://www.jlcomp.demon.co.uk/faq/ind_faq.html Bernard Peek wrote in message ...Received on Thu May 23 2002 - 19:57:46 CEST
>Pardon the hyperbole. There are costs to using surrogate keys. Each one
>increases the complexity of the system which necessarily increases the
>risk of bugs. Any database that uses a surrogate key when there is a
>suitable natural key has been designed sub-optimally.