Archives of the TeradataForum

Message Posted: Fri, 19 Jan 2001 @ 14:49:12 GMT

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

Don't be too quick to accept the HASHROW approach. It's possible for HASHROW to return a hash synonym (the same hash value for different input values). So even if the combination of (Setl_Dt, Bas_Artl_Nbr, Bsns_Lctn_ID) is unique, you can't be assured that the result of HASHROW will be truly unique.

If the primary index for your source and target table is (Setl_Dt, Bas_Artl_Nbr, Bsns_Lctn_ID), then it's not surprising that HASHROW did not cause data re-distribution. After all, the distribution of data is controlled by the hash value of the PI and the HASHROW function returns that hash value.

Remember that internally RowID is added to the PI hash value to make each row to resolve hash synonyms (and make each row truly unique). To assure uniqueness of each row that you'll be INSERTing into the target table, then you'll have to add some other value to the HASHROW result. The moment you do that, you'll have data re-distribution.

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