Archives of the TeradataForum

Message Posted: Fri, 19 Nov 2004 @ 15:38:36 GMT

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

There is a Pertner's customer presentation from 2003 that shows some empirical testing. The one issue was whether or not to convert VARCHAR to CHAR and then COMPRESS. There are tradeoffs in that case, but tends to generalize to the larger the difference between the VARCHAR average length and maximum then the less benefit (or more negativity) achieved.

In 4 empirical assessments with MVC we have seen 26 - 34% space saving (which has a direct positive impact on the initial row qualification/spooling of the compressed table(s)). the one interesting case we had was when the initial table chosen for MVC analysis was a demographic table with 158 CHAR(1) columns. R4 version had single value compression and the combination of "column value demographics and MVC compression bit overhead" actually caused the table size to grow by 26%. But that was the particular situation with 158 X 1 byte columns.

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