Archives of the TeradataForum

Message Posted: Tue, 18 Feb 2003 @ 08:53:30 GMT

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

Frank,

On the "very quick load", it is the load (of the target table) which is quick and not the read (of the source tables). If you look at the explain of the MSR (how about that for old-fashioned?) you'll see that the two reads are done sequentially.

The optimiser will apply the same logic in your "one into two" example below.

You won't get sync scan on a single query because that requires two or more separate queries. If you ran the two INS/SEL requests as separate requests then you may get sync scan. Note that the two reads of the source table (steps % and 6.2) are "eligible for sync scan". Also, I think that although the optimiser will mark the steps as eligible for sync scan it is the file system (i.e. at each AMP) which makes the decision as to whether or not the sync scan occurs.

Cheers,

Dave

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