Archives of the TeradataForum

Message Posted: Wed, 25 Jul 2007 @ 14:03:24 GMT

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

<-- Anonymously Posted: Wednesday, July 25, 2007 07:53 -->

The "low confidence" earlier in the explain is likely because you have some stats (e.g. on the PI) that tell the optimizer how many rows are in the table or spool you are starting with but not enough info regarding the columns and values involved in the predicates (ON or WHERE) to know what fraction of the rows will actually be placed in the new spool.

If the predicates use constant values, collect stats on the corresponding column or combination of columns to improve the confidence level. If the values come from a JOIN, "low" may be the best you can do. If the estimated cardinality is in the right ballpark, don't worry too much about confidence.

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