Archives of the TeradataForum

Message Posted: Wed, 26 Apr 2006 @ 18:08:15 GMT

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

<-- Anonymously Posted: Wednesday, April 26, 2006 13:05 -->

You partially answered your own question:

"cursor based row by row processing in SPs fail[s] to leverage ... parallellism effectively"

That's the primary reason to avoid cursor processing as much as possible in a parallel DBMS implementation. But you could certainly choose to put INSERT/SELECT statements in a SP rather than BTEQ script.

Sometimes there is a tendency to do many simple transformation steps instead of a few more complex steps. That's an implementation choice - tradeoff between simplicity and performance.

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