Comments on: Everything Changes

By: Ben Prusinski

Ben Prusinski — Fri, 25 Apr 2008 20:12:24 +0000

Greetings Jonathan,

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

By: Log Buffer #89: a Carnival of the Vanities for DBAs

Log Buffer #89: a Carnival of the Vanities for DBAs — Fri, 21 Mar 2008 17:00:29 +0000

[...] is Oracle’s ever-evolving Cost Based Optimizer. Jonathan Lewis at Oracle Scratchpad looks at CBO and its erratic growing pains from version 9 onwards and why it behaves the way it [...]

By: Index not used (10g) « Oracle Scratchpad

Index not used (10g) « Oracle Scratchpad — Thu, 13 Mar 2008 13:33:46 +0000

[...] by Everything Changes « Oracle Scratchpad — March 13, 2008 @ 1:31 pm UTC Mar [...]

By: Jonathan Lewis

Jonathan Lewis — Wed, 12 Mar 2008 20:23:35 +0000

Alberto,
There are no obvious clues in the 10g trace, but the 11g trace has a comment about ‘Column Groups’ - so the last change may simply be a part of the general enhancement relating to the “extended statistics” (the reference to PredCnt 3 is explained by the actual code I used - see below):
ColGroup Usage:: PredCnt: 3 Matches Full: #0 Partial: Sel: 0.0500

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 ;


create index t1_i1 on t1(n1, ind_pad, n2)

pctfree 91

;

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

By: Scott

Scott — Wed, 12 Mar 2008 19:39:05 +0000

I’m not able to reproduce this behavior on my 10.2.0.3.1 instance. Would you please post your table and index create scripts?

By: Alberto Dell'Era

Alberto Dell'Era — Tue, 11 Mar 2008 22:44:42 +0000

I have a curiosity if you don’t mind and have some time to “spare” - does the CBO using distinct_keys show up in the 10053 trace as something like “using concatenated index cardinality”, or is this improvement used silently ?