Re: Normalization, Natural Keys, Surrogate Keys

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Thu, 23 May 2002 18:57:46 +0100
Message-ID: <1022176597.13040.0.nnrp-01.9e984b29_at_news.demon.co.uk>


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 ...

>
>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.
>
Received on Thu May 23 2002 - 19:57:46 CEST

Original text of this message