Archives of the TeradataForum

Message Posted: Mon, 06 Mar 2006 @ 10:55:04 GMT

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

Hello,

As many have stated already, a Fastload is faster than an equivalent MultiLoad. For a long time I've wondered why this is so, and I don't think anyone has argued why it is so?

Previously someone tried to convince me by the "fact", that MLOAD uses a worktable and Fastload don't.

Hmmm - how can Fastload be restart able, if it doesn't use a worktable?

A couple of days ago, I believe that I figured out the reason for the performance diff. It's true, that Fastload don't specify an explicit worktable, but the thing is, that Fastload uses the target-table itself as "a kind of worktable"!

At the end of phase 1, this "table" is UNSORTED - I guess it's just a bunch of data blocks inaccessible by SQL.

Where as the MLOAD worktable is a "normal" sorted table.

So, my 2 cents is, that the performance difference equates to the time it takes to sort the input-table.

Ole Dunweber
Coop Nordic

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