Hello everyone. I think I need some guidance to get my query to use an
index. When I look at a plan table for a query I see "TABLE ACCESS
FULL" with bytes 186M.

-----------------------------------------------------------------------------------

(%CPU)| Time     |
-----------------------------------------------------------------------------------

(5)| 00:00:34 |

(5)| 00:00:34 |

(0)| 00:00:01 |

(3)| 00:00:33 |
-----------------------------------------------------------------------------------

The query performance is okay but I expected to see some index use
rather than "TABLE ACCESS FULL". Alternatively I don't understand what
the plan table is telling me so I would appreciate you telling too if
this is the case.

Table tbl_fval contains about 5million rows and table tbl_fvaljoin
contains a lot less (only contains 3 rows in the sample data below but
in our production system is contains a lot more). We are using Oracle
10.1.0.4 on Windows 2000 SP2. I have included the ddl to create the
tables, indexes etc, and the SQL to use the plan table. Am I missing
something to get the cost down (and reduce the 186M above)?

Thank you
Barry

create table tbl_fval(b_id number,b_fid number,b_fval number);
--
--and now insert lots and lots of data
--
create index idx_fval_id
on tbl_fval(b_id);
create index idx_fval_fid
on tbl_fval(b_fid);
create index idx_fval_fval
on tbl_fval(b_fval);

create table tbl_fvaljoin(b_fid number,b_fval number);
insert into tbl_fvaljoin values(100,10);
insert into tbl_fvaljoin values(101,12);
insert into tbl_fvaljoin values(103,150);

create index idx_fvaljoin_fid
on tbl_fvaljoin(b_fid);
create index idx_fvaljoin_fval
on tbl_fvaljoin(b_fval);

explain plan for
select t1.b_id
from tbl_fval t1,tbl_fvaljoin t2
 where t1.b_fid=t2.b_fid
   and t1.b_fval=t2.b_fval;

--
--this is the plan table shown at the start of the message
select plan_table_output from table(dbms_xplan.display);