Archives of the TeradataForum

Message Posted: Sun, 09 Mar 2003 @ 12:32:12 GMT

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

Dieter has pointed out most of the options, but I'd like to clarify a couple of them.

Using the Multiload IMPORT task is probably what you'll need if your delete criteria is "complex". To use this approach, run a SELECT first to identify the PI and Primary Key of every row that needs to be deleted and then export this answer set to a file on your host system. In your existing Multiload IMPORT job inserts the new rows, add a second IMPORT which reads this new file and passes the rows to a new .DML statement which issues the delete.

Another way of deleting large numbers of rows is to insert/select to a new table, but that requires disk space for the copy of the rows that you're keeping (in your case @1.17 billion). So whilst a quick way of doing it, practicalities may prevent you from taking this approach.

Dave

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