Archives of the TeradataForum

Message Posted: Sat, 06 Oct 2001 @ 15:47:15 GMT

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

The plan is the best one Teradata can do. ni_cktid is non-unique and the given ni_chtid can result in many different rows - here estimated as 16. These rows are redistributed by clli_cd to all amps and than in step 3 joined to the location table. For your suggestion you need unique primary indexes on both tables.

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