Archives of the TeradataForum

Message Posted: Thu, 07 Apr 2005 @ 15:56:51 GMT

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

Take a closer look at your stats.

The explains indicate:

"High confidence" that the prdpos.pos_txn fact table has only 28 non-NULL rows.

If table is that small, DPE isn't worth the overhead.

If table is much larger, stats are invalid and need to be updated. "No confidence" as to how many per_id's will be retrieved for a narrow range of cal_date's.

Indicates you probably need to collect stats on PRDPOS.ws_dim_time.cal_date (and keep stats up to date if the dimension table grows).

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