Archives of the TeradataForum

Message Posted: Mon, 04 Jun 2001 @ 07:43:57 GMT

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

We were facing a similar dilema about a year ago. What we finally did is, that within the transformation process we denormalize the sign information into a new column (Tran_Sign) of BYTEINT type (+1 or -1). The transaction amount is saved as it comes (ie. hopefully positive). We store the tranasaction type too.

Our model is otherwise almost normalized. Besides the sign we denormalize only the account owner ID and the account currency. Like this we avoid the most obvious joins on our largest table which really helps (the transaction table has 100 GB and we are on two nodes of WM 3700).

Petr Horsky

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