Archives of the TeradataForum

Message Posted: Fri, 08 Nov 2003 @ 00:27:07 GMT

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

I sure would like to see that explain. What was the cardinality estimate for the duplicated spool? Was it equal to your number of amps, or equal to your number of amps times 3.5 million?

In the step where your 3.5 million row table was duplicated, were there any terms to eliminate rows where any of the index columns were null? The optimizer often adds such terms in order to get rid of rows it knows will not be joined later on.

I'm a bit confused. Are you joining your 3.5 million row table on its primary index? What is the point of that? If at least one of the PI columns is null in every row, you know in advance that the operation will return no rows. What am I missing here? Are you using terms like (t1.x is null and t2.x is null) or (t1.x = t2.x) ?

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