Re: AND_EQUAL

Richard Foote wrote:

> If either index1 has not been analyzed or if index1 is very small (say
> heaps of nulls), then the optimizer might be thinking, "I might just as
> well read all the entries in this small index as it's inexpensive to do so
> and then determine the condition for my count". This could be a cheaper
> alternative than having to also read the relatively expensive index2.

All three indexes were without stats (I did ran an analyze this weekend, but have not yet had time to re-visit this issue).

As for nulls - no nulls.

Thanks for the reply though. Usually my questions here are met with a spontanous round of indifference. :-)

--
Billy