Archives of the TeradataForum

Message Posted: Tue, 30 Nov 2004 @ 20:31:26 GMT

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

I've done CRM testing on a retail data model about 3 years ago. I tested CRM queries that went against each of the following:

1. De-normalize the header and detail table into single table

2. Use normalized model with DATE, STORE, CUST_ID as the PI for both header and detail.

3. Use normalized model with basket_id as PI for both header and detail.

Of the three solution, option 3 performed the best in regards to query time, cpu utilization, and spool consumtion.

Hope this helps....

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