Archives of the TeradataForum

Message Posted: Fri, 14 Mar 2003 @ 11:09:36 GMT

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

I ran a test and I found unexpected results favoring mload delete over insert-select.

wall time in secs (mload = 75, insert-select = 2160)
AMP CPU (mload = 1686, insert-select = 39067)
Disk I/O (mload = 4305018, insert-select = 14124056)

The table had 500 mill rows and I was deleting 100 mill (20%) on a WHERE on a non-PI column date column. There are no SIs, and there were no blocking during the runs. I am using V2R4.1

What could explain the above results?

Kumaran Anantaraman

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