Archives of the TeradataForum

Message Posted: Tue, 11 Oct 2005 @ 13:00:51 GMT

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

Hello,

I frequently see the same behavior ever since we moved to V2R5 and now V2R6. We also have a high number of AMPs (478) and a small table with all our stores (about 600 rows).

Joining it to a big fact table results in the fact table being moved into a Spool and redistributed.

This always takes a lot of time.

The solution Dieter suggested (add 0 to a number or concatenate '' to a string) in the WHERE-clause usually works fine.

The optimizer then duplicates the small table and joins it to the big table, saving on Spool and execution time.

I'm sure changing the PI on the small table also works fine, too, but usually that will have negative impacts on other steps.

Hope that helps.

Roland

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