Archives of the TeradataForum

Message Posted: Fri, 26 Mar 2004 @ 17:08:51 GMT

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

I have a question regarding an aggregate join index defined with fallback.

I have a detailed transaction table (non-fallback), to which I intend to add an aggregate single-table join index with fallback. This aggregate is basically my daily sales summary. Now I create a view called DailySales, using the exact SELECT statement defined in the join index. Whenever the end user SELECTs from the DailySales view, the optimizer allways chooses the join index, great!

But now an amp goes down. My transaction table is unavailable. Is my join index available?

My question is this - Is there anything going on in the parser/optimizer that would cause it to need to reference the unavailable table? Something in the table header, perhaps, like some presence bits indicating a defined dependent join index? Or maybe the optimizer needs to compare Random-Amp sampled statistics to know if the join index substitution makes sense?

Obviously, I cannot simulate the amp-down condition, so I am wondering if anyone can help/explain.

Thanks

Donald

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