Archives of the TeradataForum

Message Posted: Wed, 29 Oct 2015 @ 00:37:31 GMT

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

The duplicate row checking is still done, but with a UPI you only have to check at most one row so the overhead is tiny. With NUPI, the overhead can be substantial even with a USI.

If you have a SET table with a UPI or USI, then multiple-row operations (e.g. INSERT SELECT or MERGE) will quietly eliminate duplicate rows , whereas with MULTISET table and UPI or USI, the duplicate key check will cause the operation to fail. For single-row operations (e.g. INSERT VALUES), SET/MULTISET both give duplicate key error.

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