Your observation is correct – this cost is more realistic, and the change is an improvement in the optimizer’s algorithms.

The point I was trying to make, though, was that any change (even a logical improvement) will be bad news for someone because any change in an execution path may be unlucky for there particular case; so knowing about the existence of the change, and some details of how the change works, is important.

Is it not true that in 11.1 the Optimizer sees a more realistic value of the index & table selectivity ? Given this, is this plan not better than the ones in 10g. The 10g plans seem to have a lesser cost but they don’t seem to have calculated the selectivity correctly ?

Please evaluate my understanding.

Thanks for the excellent tip on how the execution plans display differently between Oracle 10gR1 and Oracle 11gR1. By the way hope to see you in San Diego sometime in the future.

Cheers,
Ben Prusinski, Oracle DBA/Consultant

Scott,
The SQL, excluding the call to dbms_stats, is as follows:

create table t1
as
select
        mod(rownum,20)          n1,
        mod(rownum,20)          n2,
        rpad('x',40,'x')        ind_pad,
        rpad('x',200)           padding
from
        all_objects
where
        rownum <= 4000
;

select
        /*+ 
                index(t1) 
        */
        padding
from
        t1
where
        n1      = 2
and     ind_pad = rpad('x',40,'x')
and     n2      = 2
;