Archives of the TeradataForum

Message Posted: Mon, 03 Jun 2002 @ 22:39:56 GMT

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

I'm not aware of any, but there are come gotcha's namely:

The following sequence isn't an insert into an empty table:

delete from tn
;insert into tn
;insert into tn

While the following is:

delete from tn;
insert into tn
;insert into tn

This one isn't:

insert into tn
;
select from xyz;
insert into tn;

The semicolon has to immediately precede the next insert in BTEQ.

Also, sending multiple inserts into an empty table as a parallel request in ODBC(Queryman) isn't an insert into an empty table.

These subtle differences are probably why your support guy may feel that journaling is done on inserts into empty tables.

A large (tens of thousands) of PI duplicates are a much more likely candidate for any slowdowns that you might see. Try making the target table a multiset table and do something after the load to detect and suppress duplicates.

Pat Belle
Teradata Certified Master

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