Archives of the TeradataForum

Message Posted: Wed, 15 Dec 2011 @ 01:06:03 GMT

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

Here is what happens:

When you define a multi column PI sata distribution will be perfect. Which is good . But for end users it doesn't matter how the data was distributed but how fast the data has been retrieved. Having said that always users won't query with all the multi columb where clause and multi column joins .

So if your data has been distributed on multi column pi and if one of the columns is being used to join then Teradata has to redistribute your table on that column ( whether it is local to amp or duplicated on all amps) depends on other factors. So the multicolumn pi is not at all helpng on this scenario.

Hope it helps.

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