Archives of the TeradataForum

Message Posted: Fri, 20 Apr 2007 @ 21:46:36 GMT

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

When doing MV compression analysis, I have to ultimately run a query for each column within a table. So for an application with 200 tables, averaging 15 columns per table, I find myself generating and running a lot of queries. (The query generation is not the issue - this is automated; the issue is that every attribute needs its own analytical pass thru the data.)

Has anybody seen (or better is using) a method that makes only one pass on a table to perform its MV compression analysis? In other words, gather all the frequencies of occurrence for each of the big hitters on each column of the table - AT ONCE. I'd settle for something that identifies the top 7 or top 15 for each column.

Another approach would pull the vital info from dbc.indexstats, columnstats, multicolumnstats. Has anybody tried that? (Maybe everything needed is not there?)

Along the lines of the DBC Stat tables, has anybody seen a layout for the IndexStatistics, FieldStatistics, ColumnsStatistics.

I have not been able to find a layout of any of these tables in the Teradata documentation.

It's Friday here which means the weekend, so for everyone heading home for the weekend, have a great weekend.

Thanks,

Bill Gregg
Teradata Team
Thomson Healthcare

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