Archives of the TeradataForum

Message Posted: Thu, 07 Apr 2005 @ 16:21:16 GMT

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

Wayne,

Is it possible that the behavior has anything to do with either the size of the table, statistics, or the fact that per_id is NULLABLE?

I notice that in example 1, the optimizer reports a High Confidence estimate of 28 rows.

You don't specify what statistics were collected. The high confidence tells me that you probably have stats on PER_ID in the transaction table.

I notice that the explains are eliminating all NULL PER_IDs in the extraction steps.

I am not claiming any special knowledge. We are just starting testing and new feature evaluation on 5.1.2. There is a lot of interest in this feature. Particularly when we move to V2R6.0 and Backups by partition.

I would be extremely interested in finding out what to avoid.

Thanks,

Glen

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