Stefan,

> > Please create a new compound index based on Word_1.attribute and
> > Word_1.data (in that order) (or is IXWORD3 that index?).

> Hello Sean,
>
> THANKS!!!!!

Actually, Anderson Farias deserves credit as well. Without his
narrowing of the problem to a simple SQL which we could focus on, there
would have been no way I would have bothered to work further on problem.

> Creating the Index IdxWordtest on word (attribute, data) did the job.

It was the uniqueness of the resulting set which suggested that the
compound index would be beneficial.

BTW, that is the definition of the IXWORD3 index?

If it is on Word (attribute), then it can be dropped. The new index can
be used instead for searches on Attribute alone as well as Attribute +
Value.

> Both queries runs in about 3-4 seconds.

That is much faster than I had expected, I expected an improvement of
only 5-20x times.


Sean

P.S. Like SET I still think that the BOLD tool needs to review how it
constructs its SQL queries -- WHAT A MESS!