Archives of the TeradataForum
Message Posted: Sun, 02 Mar 2003 @ 14:22:16 GMT
Subj: | | Re: Performance for Distinct values |
|
From: | | David Wellman |
The second point that you refer to here is irrelevant.
In an insert/select the select is treated as a sub-query. Teradata never returns duplicate rows from a sub-query (note that a derived
table is not a sub-query), hence the de-duplication is done prior to the Insert phase, which means that Set/Multiset is irrelevant. When
you don't code either a GROUP BY or a DISTINCT Teradata will use DISTINCT processing logic.
It is the other factors which determine whether the GROUP BY or DISTINCT processing goes faster -- i.e. the number of rows/value.
Dave
|