Archives of the TeradataForum

Message Posted: Thu, 11 Dec 2008 @ 20:28:44 GMT

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

Varun:

Since the PI of the small table is the join column, then only way to avoid a product join would be to redistribute by hash code all the rows of the large table. I don't know about you, but this does not sound like a good idea to me. Therefore, duplicating all the rows of the small table onto each AMP and doing a product join (which by the way requires no sorting) in parallel on all AMPs at the same time seems to be the winner as far performance goes.

Are you unhappy with performance or the fact that you see "product join" in the EXPLAIN?

Regards,

Michael Larkins
Certified Teradata Master
Certified Teradata SQL Instructor

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