Archives of the TeradataForum

Message Posted: Tue, 25 Oct 2005 @ 09:03:08 GMT

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

I am certainly missing something here....

1) I got the impression that the extract effort is the same as the select part in the select insert statemet except for the actual IO to a file system.

With this assumption, we only have to look at the benefits of mload/fastload.

Fastload my have a benefit if we load into an empty table. This can be done in blocks which certainly is fast. Why this cannot be done with the select insert seems more of an implementation question than a technical requirement. I guess in this case you may get a benefit of fastload if the latter makes up for the additional IO to extract to and load from the filesystem.

On the mload case, you conceptually run a fastload into a worktable followed by a select insert into the final table. Not sure how this ever could be faster than an original select into.

Does anybody have quantitative numbers on this?

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