On Jun 28, 2008, at 12:53 AM, Bob Duffey wrote:
>
> If you're iterating through the records with a cursor, the plan may
> be different, IIRC - weighted to provide first row quickly, as opposed
> to the query that was run that's weighted to provide last row quickly.
>
> I agree, and I was hoping that would be the case, but as it happens
> it wasn't. Anyway, reducing random_page_cost seems to have resulted
> in the "right" plan being selected.
>
The original query you mentioned was not using a cursor, though, it
was just a select.
Did you try it using a cursor?
Cheers,
Steve