| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> RE: oracle can ignore hints
My example is a little bit poor. I have seen bad plans created from Oracle
accepting a hint that should not have been used, at least not in isolation.
Normally, the main cause of Oracle ignoring INDEX hints from what I have
seen is the CBO using a hash join. To get rid of the hash join I order the
tables in the FROM clause from left to right and use the ORDERED hint plus a
USE_NL. This will normally change the access path from a hash join to a
nested loop join and allow use of the indexes, generally without the index
hint being required.
The way I see it is the join order and method pretty much dictate which indexes are possible for use. The table A, table B example was a poor choice.
-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org]On Behalf Of Jonathan Lewis
Sent: Friday, March 05, 2004 7:30 AM
To: oracle-l_at_freelists.org
Subject: Re: oracle can ignore hints
Notes in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
March 2004 Hotsos Symposium - The Burden of Proof
Dynamic Sampling - an investigation
March 2004 Charlotte OUG (www.cltoug.org) CBO Tutorial
April 2004 Iceland
June 2004 UK - Optimising Oracle Seminar
: Based on experience for versions 7 - 8 the CBO will do as it is told via a
: syntactically valid hint if the join order and method allow do not
: invalidate the hint. If you provide an index hint to use an index into
: table B and Oracle chooses to drive on table B then the hint is unusable.
I don't think that's true, but I don't have a 7 or 8 on hand to check. The script at the end of the note shows it to be false in 9; but more significantly shows how Oracle will follow a totally ridiculous path if hinted correctly.
drop table t1;
create table t1 (
id number,
v1 number,
small_vc varchar2(10),
padding varchar2(100)
);
drop table t2;
create table t2(
id number,
v1 number,
small_vc varchar2(10),
padding varchar2(100)
);
insert into t1
select rownum, 1, rpad('x',10), rpad('x',100)
from all_objects
where rownum <= 5000;
commit;
insert into t2
select rownum, rownum, rpad('x',10), rpad('x',100)
from all_objects
where rownum <= 5000;
commit;
alter table t1 add constraint t1_pk primary key(id); alter table t2 add constraint t2_pk primary key(id);
create index t1_junk on t1(v1);
analyze table t1 compute statistics;
analyze table t2 compute statistics;
set autotrace traceonly explain
spool temp
select t1.small_vc, t2.small_vc
from
t1,t2
where
t2.id = 99
and t1.id = t2.id
and t1.v1 = 1
;
select
/*+ ordered index(t1 t1_junk) */
t1.small_vc, t2.small_vc
from
t1,t2
where
t2.id = 99
and t1.id = t2.id
and t1.v1 = 1
;
spool off
doc
Execution plans on 9.2.0.4 with DBCA installed database CBO chose this path:
Execution Plan
0 SELECT STATEMENT Optimizer=CHOOSE (Cost=4 Card=1 Bytes=28) 1 0 NESTED LOOPS (Cost=4 Card=1 Bytes=28)
2 1 TABLE ACCESS (BY INDEX ROWID) OF 'T2' (Cost=3 Card=1 Bytes=13) 3 2 INDEX (UNIQUE SCAN) OF 'T2_PK' (UNIQUE) (Cost=2 Card=5000) 4 1 TABLE ACCESS (BY INDEX ROWID) OF 'T1' (Cost=2 Card=1 Bytes=15) 5 4 INDEX (UNIQUE SCAN) OF 'T1_PK' (UNIQUE) (Cost=1 Card=5000)
Elapsed: 00:00:00.00
CBO did as it was told - and got all 5,000 rows from the T1 table using a completely useless index.
Execution Plan
0 SELECT STATEMENT Optimizer=CHOOSE (Cost=99 Card=1 Bytes=28) 1 0 MERGE JOIN (CARTESIAN) (Cost=99 Card=1 Bytes=28) 2 1 TABLE ACCESS (BY INDEX ROWID) OF 'T1' (Cost=98 Card=1 Bytes=15)
3 2 INDEX (RANGE SCAN) OF 'T1_JUNK' (NON-UNIQUE) (Cost=11 Card=5000)
4 1 BUFFER (SORT) (Cost=1 Card=1 Bytes=13) 5 4 TABLE ACCESS (BY INDEX ROWID) OF 'T2' (Cost=2 Card=1 Bytes=13) 6 5 INDEX (UNIQUE SCAN) OF 'T2_PK' (UNIQUE) (Cost=1 Card=5000)
#
-- Archives are at http://www.freelists.org/archives/oracle-l/ FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html ----------------------------------------------------------------- ---------------------------------------------------------------- Please see the official ORACLE-L FAQ: http://www.orafaq.com ---------------------------------------------------------------- To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line. -- Archives are at http://www.freelists.org/archives/oracle-l/ FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html -----------------------------------------------------------------Received on Fri Mar 05 2004 - 10:25:55 CST
 |
 |