Re: reference cursors

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Fri, 30 Jan 2004 16:44:59 -0000
Message-ID: <006601c3e750$6d6ac6a0$6702a8c0@Primary>

According to the 9.2 manual, the cardinality() hint should take a list of table aliases, and a cardinality. Then when that list of tables starts the join order, the cardinality of the join up to that point is what you have specified.

Consequently the four separate cardinality hints in your query are saying:

    if you start with t1, the cardinality from t1 is 10
    if you start with t2, the cardinality from t2 is 10
    if you start with t3, the cardinality from t3 is 10
    if you start with t4, the cardinality from t4 is 10

This may not be what you intend. As a consequence, it is possible that some other small effect relating to ref cursors or dynamic sql is causing a different table ordering, and a dramatic change in the statistics that Oracle is required to use.

L:ine 11 may be a big part of your problem. Oracle assumes the cardinality of a table cast() is 8,168 - but your run time stats show 2 rows returned.

Regards

Jonathan Lewis
http://www.jlcomp.demon.co.uk

  The educated person is not the person
  who can answer the questions, but the
  person who can question the answers -- T. Schick Jr

Next public appearances:
Jan 29th 2004 UKOUG Unix SIG - v$ and x$ March 2004 Hotsos Symposium - The Burden of Proof March 2004 Charlotte NC OUG - CBO Tutorial April 2004 Iceland

One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html

Three-day seminar:
see http://www.jlcomp.demon.co.uk/seminar.html ____UK___February
____UK___June

The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html

Original Message ----- From: <Stephen.Lee_at_DTAG.Com> To: <oracle-l_at_freelists.org> Sent: Friday, January 30, 2004 2:59 PM Subject: RE: reference cursors

Interesting suggestion. I might fiddle with it some more. Here is what is in the 10046 level 12 trace file.

PARSING IN CURSOR #12 len=2181 dep=1 uid=19 oct=3 lid=19 tim=16595183616 hv=1672063353 ad='117f50f0'
SELECT /*+ NO_EXPAND INDEX (tmr TMR_LCT_FK_IDX) INDEX (tmri TMRC_MIN_DY_IDX) cardinality (t4, 10) cardinality (t3, 10) cardinality (t2, 10) cardinality (t1, 10) */ etc. etc.

And one can clearly see where the thing hits the wall: STAT #12 id=5 cnt=13180531 pid=4 pos=1 obj=0 op='NESTED LOOPS (cr=25497 r=4 w=0 time=30502912 us)'
...
...

STAT #12 id=14 cnt=13179320 pid=5 pos=2 obj=223485 op='INDEX RANGE SCAN TMRC_MIN_DY_IDX (cr=24202 r=0 w=0 time=20999168 us)'

When the hints work, the query does very little work.

You DO have to wonder just exactly what is a COLLECTION ITERATOR PICKLER FETCH. Is that a kosher pickler?

STAT #12 id=11 cnt=2 pid=10 pos=1 obj=0 op='COLLECTION ITERATOR PICKLER FETCH (cr=0 r=0 w=0 time=0 us)'

-----Original Message-----

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 Jan 30 2004 - 10:44:59 CST