Archives of the TeradataForum

Message Posted: Wed, 18 Aug 2004 @ 23:02:33 GMT

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

You're right. If the select is a union, we should be able to bypass dupe checking for the insert. I don't know what I was thinking.

On dupe checking overhead, it depends on how many NUPI dupes there are. With lots, it's expensive. As each rows is inserted, we only have to dupe check it against other rows with the same rowhash.

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