Archives of the TeradataForum

Message Posted: Tue, 21 Feb 2006 @ 20:44:15 GMT

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

Hello Ananta,

In your query, you have not supplied all of the information (values from both columns) necessary to generate the complete hash value. It will do a table scan (assuming no other indexes are in place). So, the catch is that ALL columns in the PI must be given values in the query for the PI to be used as an access method.

I would guess that an actual value of dw_eff_ts is never quoted in a query, therefore it has only been added for index uniqueness. If having a UPI is not important, I would think that a PI of acct_id (is this a transaction table) would be quite well distributed. You could also consider partitioning.....

Andrew W

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