Archives of the TeradataForum

Message Posted: Wed, 05 Nov 2003 @ 16:54:00 GMT

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

My concern is that the optimizer (in our case) is looking at the number of unique values of the primary index in order to decide how to join this table to other tables. Because only 1 unique value exists, the optimizer decides to duplicate this table to all amps. Unfortunately, our table actually has 3.5 million records (eventhough only 1 unique primary index value exists from a 'statistics' point of view. This is not very efficient.

Perhaps, as a solution, the optimizer should look at the record count (instead of the number of unique values of the primary index) when deciding whether or not to duplicate a table to all AMPs as part of a join plan.

Seem reasonable?

Clay

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