Archives of the TeradataForum

Message Posted: Thu, 23 Sep 2004 @ 19:19:10 GMT

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

Prasanna:

Since you do not tell us which one is slower, nor you did tell us if the data contains duplicate values in the PI and you did not tell us what type of PI is on your test_insert table, we can only speculate.

The only thing we know is that table_create has a NUPI. If you need further analysis, you might provide the CREATE for test_inserter also. That is the only way we can compare apples with apples.

So, based on not having sufficient data I am guessing that the create table with the derived table is slower. I say this for two reasons: 1. thinking that test_inserter has a UPI. Therefore, it will not require any duplicate row checking like your test_create that has a NUPI, if there are duplicate values. 2. The CREATE TABLE has to insert multiple rows into the data dictionary (TVM, TVFIELDS, ACCESSRIGHTS, INDEXES) to build the table_create before inserting a single row into it.

That is my shot in the dark,

Michael Larkins
Certified Teradata Master
Certified Teradata SQL Instructor

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