Archives of the TeradataForum

Message Posted: Thu, 23 Jun 2011 @ 12:27:15 GMT

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

Thanks for your answers folks.

Barry, you're probably right. There were no statistics at all involved. However, the system 1 table had some lines in it, while system 2 table was empty. Even though COLUMN2 = 2 brings 0 lines in both cases, amp-sampling estimate thinks otherwise. Over 500 for system 1 and 1 line estimated for system 2. When I recreated the table in system 1 from the scratch and let it empty, I got the same execution plan for both.

Thanks for help.

BTW, flushing cache cannot get blocked when there is some DDL operation at the same time, like creation of index? I think I remember case like that and it would probably explain the different response times on my 2 systems.

Regards,

Tomas

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