Archives of the TeradataForum
Message Posted: Wed, 15 Jan 2003 @ 15:44:33 GMT
There is a different between a Relational product join type and a product join data access strategy. Please excuse the terms. I don't know if we really have a commonly defined set of terms to describe this.
Relational approach states that a product or cross join type will occur if you fail to include a join condition between two tables.
Teradata has many join strategies; one of which is a product join strategy. Others include a merge join (inclusion and exclusion), a hash-join, etc...
The product strategy is made up of both a data re-location (geography) strategy (cost to move the data to where it can be joined; Amp Local) and a data join (cost to compare each row with each other row in Product join or cost to sort and then compare left ordered rows to right ordered rows in merge join) strategy. These two parts are what makes up the cost basis for deciding which Teradata join strategy to use. This is part of the re-write process that occurs in Teradata.
Like others have said, if the optimizer didn't decide to use a product join strategy. Then it would have had to redistribute the large table over to the amps with the same PI as the small tables PI, sort all of the data in that order and then do a merge join.
I hope this helps,
|Copyright 2016 - All Rights Reserved|
|Last Modified: 23 Jun 2019|