Archives of the TeradataForum

Message Posted: Tue, 25 May 2004 @ 11:47:52 GMT

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

Hi,

I have enjoyed and learned a lot from reading this forum, this is my first post.

Daily we "generate" (build from scratch) a lot of aggregate tables from base-tables. Of course, performance is crucial.

The aggregation always involves a join to one or more (small) dimension tables.

Initially I didn't think I could influence the sum-step in any way. But it makes a lot of difference whether aggregates are computed locally or globally.

We achieve "locally" computation if the PI of the base- and aggregate-table are identical (and the join is local).

Questions:

1) What exactly is going on when aggregates are computed globally?

2) Will aggregates be computed locally only if PI is the same?

Ole Dunweber

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