Archives of the TeradataForum

Message Posted: Thu, 01 May 2003 @ 13:12:17 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

<-- Anonymously Posted: Thursday, May 01, 2003 09:08 -->

I apologize in advance if this topic has come up on numerous occasions. I'm relatively new both to this list and Teradata - so hopefully this question has not been posed too commonly/recently. (I don't have Internet Access yet or I'd just check the archives).

My question is of a general nature, so I hope that I am able to make the reason behind the question clear in short order.

I'm under the impression that surrogate keys are not commonly used with Teradata - which is always(?) normalized. Is this the case, and if so why? Every data warehouse I've been associated used was unnormalized (Star Schema) and used surrogate keys. At my new company, we have many varied data sources with no ability to guarantee that customer numbers, material numbers, etc are not re-used. Each source also has different "recycle" policies or different attributes (ie - material numbers may never be reused from one source, but may be somewhat regularly re-used from another). It would seem to me that surrogate keys would greatly simplify some of the complexity of this type of warehouse.

I'm assuming that most of you using Teradata are global companies in very similar situations. Do any of you use surrogate keys and why? For those who do not, how do you guarantee that your "natural" keys will not be violated over time - or do you have more centralized source systems and this is not possible in your case?

Thanks in advance for any feedback on this topic.

	<Prev	Next>		<<First	<Prev	Next>	Last>>