Archives of the TeradataForum

Message Posted: Tue, 19 Feb 2003 @ 00:27:43 GMT

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

There is no step so far in Teradata that is capable of reading a table and writing to multiple spools. Not impossible, just something we have never done. This request requires different spools because the contents are completely different (different date ranges selected). If the spools were the same, you would see the optimizer eliminate the duplicate scans and the duplicate spool.

As Dave Wellman pointed out, sync scan is intended to be inter-query, not intra-query. It is very occasionally possible to have a sync scan within a query but only when the steps occur within the same parallel steps block. This will not typically be the case.

I also think that if you broke the query apart, you would find that the scan part is a small part of the total cost of the query when the result set is large (like this case). Most of the cost in on the spooling, sorting and inserting processes.

What release was this run on? I am a little surprised to see the sort in the plan.

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