Archives of the TeradataForum

Message Posted: Wed, 24 Nov 2010 @ 09:53:31 GMT

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

Hi,

Sample stats work well if the column(s) are fairly unique, and save 90 % of your time. However, you still need the proper collect stats on the non unique column(s) to provide accurate information to the optimizer.

Last, always collect the stats on the PARTITION after table refresh (which normally take a few second only), i.e.

     COLLECT STATS ON . COLUMN (partition);

The collect stats on PARTITION also improve the accuracy of the stats extrapolation.

Hope it help.

Regards,

YB
T&O-BICC
DBS Bank Ltd

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