This is a theorical question (or maybe some of you have experiencied.

Actually, I have a script that "queries of query" in cfmx as following:
q1 = "select fil1, fil2, sum(fl3) as sfl3 from tb1 group by fl1,fl2 order by
fl1,fl2"
q2 = "select distinct fil1 as C from q1"
q3 = "select disctint fil2 as R from q1"
cfloop q2 -> cfloop q3 -> q4 = "select sfl3 from q1 where fil1 = #C# and fil2
= #R#"

So I have 1 connection to the database then it is up to cfmx to act;

Would it be faster/better/efficient if I connect to the database:
q1 = "select fil1, fil2, sum(fl3) as sfl3 from tb1 group by fl1,fl2 order by
fl1,fl2"
q2 = "select distinct fil1 as C from tb1 order by fl1"
q3 = "select disctint fil2 as R from tb1 order by fl2"
cfloop q2 -> cfloop q3 -> q4 = "select sfl3 from (select fil1, fil2, sum(fl3)
as sfl3 from tb1 where fil1 = #C# and fil2 = #R#) group by fl1,fl2 order by
fl1,fl2"

So 3 connections + multiple connection for each crosstab. In the system,
actually I can have up to 1700x1700 values (and soon I will have to multiply
this by 25*4 - so 100 table each with 1700x1700 values) to retrieves and as you
can understand it take a lot of time to display the results. I am looking for a
way to speed up this particular situation.


PS: The database is Oracle 9i