Archives of the TeradataForum

Message Posted: Fri, 21 Sep 2001 @ 12:55:14 GMT

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

Another way to look at row distribution is by hashkey value. The SQL shown below if from one of the reference manuals (the Database Design I think). The row counts are by HASHAMP, and if you look at %'s, should track to the SQL for DBC.Tablesize by Vproc. If you skewing, I would say one or more AMPs that are GT 5-10% (although this would depend on total number of AMPS) of median then you are probably having large PI synonym chaining (i.e. many rows with same hash key value). and if you see this across the tables that you are joining (assuming same PI index joins) then it causes skewing to do the joins. If hashkey skewing is not the issue then you are probably doing a lot of aggregation and/or your join criteria may be on columns that have many NULL and/or default (e.g. 0 for NUMERIC) values.

SELECT HASHAMP (HASHBUCKET (HASHROW(primary index columns ))), COUNT(*) FROM tablename GROUP BY 1 ORDER BY 1;

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