Archives of the TeradataForum

Message Posted: Fri, 20 Mar 2009 @ 10:11:48 GMT

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

Hi Simard,

Based on my understanding (and experience) you're correct in your original guess, because you're using UNION.

The UNION function means "do NOT return duplicate rows". The way that Teradata does this is to hash distribute each row on a hash of the entire row before eliminating duplicate rows. As you've seen, even when you have the same PI, because the rows are based on their ** entire ** contents, just having the same PI is irrelevant (to the UNION processing).

To avoid the redistribution what you could try is:

1) An MSR. Teradata will silently drop duplicate rows during the insert processing.

(if the above doesn't work)

2) Try multiple insert/selects from a single source table into the target one. All apart from the first one will write a TJ, but that may be better/quicker than the redistribution.

Cheers,

Dave

Ward Analytics Ltd: Information in motion ( www.ward-analytics.com )

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