Thanks ann,

but do you know how to use the MON$ table? any doc somewhere or sample to retrieve the indexed reads vs the non indexed reads ?

about the filter, the rules is an idea, the probleme is that it will take time, time and time to create/debug/maintain it !
for exemple i see that if i have a index on (LocationID, beds)
and if i do the query
select * from property where locationid='xxx' and beds >= 1 and beds <= 2 order by locationsid, beds
then it's crazy but usind the index locationid, beds (like plan property order index_locationsid_beds) is very slow !! in the oposite way using plan property order index_locationsid_beds index (index beds) is fast !!

i must find a system that will learm by himself with plan to use, like every human do ... by guessing first and trying and trying again

--- In firebird-support@yahoogroups.com, "Ann W. Harrison" <aharrison@...> wrote:
>
> svanderclock wrote:
> >
> > someone already told me but i can not found it anymore:
> > after i do a select, how to retrieve the indexed reads and non indexed reads statistics ? like ibexpert or ems firebird manager do
> >
>
> I think those are available from the MON$ pseudo-tables.
>
> Thinking about what Arno said, you might get better results if
> you used some rules about the queries you generate. For example
> if the user specifies beds > 1, don't generate a criteria for
> beds; instead, filter the results yourself.
>
> For what it's worth, PostgreSQL does use histograms to monitor
> the distribution of values in an index.
> Good luck,
>
> Ann
>