Archives of the TeradataForum

Message Posted: Tue, 11 May 2004 @ 19:30:57 GMT

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

I agree with Martin about collecting more frequently on date columns...You should collect statistics on any column that has new values added to it (and you have constraints on it). The reason is that the optimizer knows what the upper and lower boundaries are for a given column (with statistics). If you write a query that has 'where col_date = yesterday' and you only have statistics current from two weeks ago, the optimizer assumes that the stats are stale and does a random amp sample. Depending on the data block sampled, this may or may be a good thing (the optimizer may choose a very poor path).

Ruth Fenwick
Database Services

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