Archives of the TeradataForum

Message Posted: Wed, 21 Mar 2001 @ 09:16:29 GMT

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

David,

We have compress on many of our fact columns, for example on 40 columns of our 56 column main order table. This came out of experiments for performance on a very large but sparsley populated main fact table used for a Business Objects development. Whilst there is a minor overhead to populate the table, we found a big improvement in space saving and also in query performance. I understand that this is because more data is brought into each (64k?) read block as the compressed values are much smaller. This only works if you have a value that is very common in the column in question. In our case as it is a sparsley populated dimension table where zero is very common.

Dave

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