Archives of the TeradataForum

Message Posted: Sun, 20 Aug 2006 @ 14:46:31 GMT

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

Chandana:

That is correct - Teradata has always stored rows in a table based on the Row Hash(32bits) + Uniqueness Value(32bits) = Rowid of the primary index(PI). Then when we make a query referencing the columns of the PI, Teradata can re-create the Row Hash and quickly find the row(s) you are looking for in the table on a single AMP.

With PPI as an option now, when the partition value(s) is provided in the WHERE, Teradata can still find the row with the partition using the row hash on a single AMP using the data values of the PI within the partition(s).

Hope this helps,

Michael Larkins
Certified Teradata Master
Certified Teradata SQL Instructor

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