CBO bug? Lack of SORT UNIQUE.

Question

Hi all,
Let's consider following case:
create table tmp as select rownum id, 0 sign from dual connect by level <= 100;
create index tmp_i on tmp(id,sign);
create table t as
select mod(rownum,2) id, mod(rownum,3) val
from dual
connect by level <= 100000;
begin
   dbms_stats.gather_table_stats (
      user,
      'T',
      estimate_percent   => null,
      method_opt         => 'FOR ALL COLUMNS SIZE SKEWONLY',
      cascade            => true
end;
begin
   dbms_stats.gather_table_stats (
      user,
      'TMP',
      estimate_percent   => null,
      method_opt         => 'FOR ALL COLUMNS SIZE SKEWONLY',
      cascade            => true
end;
/As you can see it scans TMP_I 50 000 times for statement with max (irrespective of distinct in subquery).
Is there any way to enforce CBO to make SORT UNIQUE for max as well as for count so that it scans TMP_I only 3 times?
SQL> select --+ leading(t) use_nl(t tmp)
  2  max(id)
  3  from tmp tmp
  4  where tmp.sign = 0
  5  and tmp.id in (select val from t where id = 1);
   MAX(ID)
         2
| Id  | Operation           | Name  | Starts | E-Rows | A-Rows |   A-Time   | Buffers |
|   0 | SELECT STATEMENT    |       |      1 |        |      1 |00:00:00.14 |     159 |
|   1 |  SORT AGGREGATE     |       |      1 |      1 |      1 |00:00:00.14 |     159 |
|   2 |   NESTED LOOPS      |       |      1 |  49750 |  33333 |00:00:00.13 |     159 |
|*  3 |    TABLE ACCESS FULL| T     |      1 |  50000 |  50000 |00:00:00.02 |     156 |
|*  4 |    INDEX RANGE SCAN | TMP_I |  50000 |      1 |  33333 |00:00:00.07 |       3 |
Predicate Information (identified by operation id):
   3 - filter("ID"=1)
   4 - access("TMP"."ID"="VAL" AND "TMP"."SIGN"=0)
SQL> select --+ leading(t) use_nl(t tmp)
  2  max(id)
  3  from tmp tmp
  4  where tmp.sign = 0
  5  and tmp.id in (select distinct val from t where id = 1);
   MAX(ID)
         2
| Id  | Operation           | Name  | Starts | E-Rows | A-Rows |   A-Time   | Buffers |
|   0 | SELECT STATEMENT    |       |      1 |        |      1 |00:00:00.14 |     159 |
|   1 |  SORT AGGREGATE     |       |      1 |      1 |      1 |00:00:00.14 |     159 |
|   2 |   NESTED LOOPS      |       |      1 |  49750 |  33333 |00:00:00.13 |     159 |
|*  3 |    TABLE ACCESS FULL| T     |      1 |  50000 |  50000 |00:00:00.01 |     156 |
|*  4 |    INDEX RANGE SCAN | TMP_I |  50000 |      1 |  33333 |00:00:00.07 |       3 |
Predicate Information (identified by operation id):
   3 - filter("ID"=1)
   4 - access("TMP"."ID"="VAL" AND "TMP"."SIGN"=0)
SQL> select --+ leading(t) use_nl(t tmp)
  2  count(id)
  3  from tmp tmp
  4  where tmp.sign = 0
  5  and tmp.id in (select val from t where id = 1);
COUNT(ID)
         2
| Id  | Operation            | Name  | Starts | E-Rows | A-Rows |   A-Time   | Buffers |  OMem |  1Mem | Used-Mem |
|   0 | SELECT STATEMENT     |       |      1 |        |      1 |00:00:00.03 |     158 |       |       |          |
|   1 |  SORT AGGREGATE      |       |      1 |      1 |      1 |00:00:00.03 |     158 |       |       |          |
|   2 |   NESTED LOOPS       |       |      1 |      3 |      2 |00:00:00.03 |     158 |       |       |          |
|   3 |    SORT UNIQUE       |       |      1 |  50000 |      3 |00:00:00.03 |     156 |  2048 |  2048 | 2048  (0)|
|*  4 |     TABLE ACCESS FULL| T     |      1 |  50000 |  50000 |00:00:00.01 |     156 |       |       |          |
|*  5 |    INDEX RANGE SCAN  | TMP_I |      3 |      1 |      2 |00:00:00.01 |       2 |       |       |          |
Predicate Information (identified by operation id):
   4 - filter("ID"=1)
   5 - access("TMP"."ID"="VAL" AND "TMP"."SIGN"=0)I can't figure out why SORT UNIQUE is absent for statement with max.
PS. 11gR2

Answer 1

Thanks for reply.
user503699 wrote:
I don't think it is a good idea to compare your last and first query timings as they are 2 different queries.Ok. I could compare query with max(id), sign(count(*)+1) vs max(id), sign(1). They always produce the same results so can be considered as the same.
But I think that max(id), count(*) vs max(id) was enough to explain my point.
user503699 wrote:
If you are so sure of that you can write something like following :
SQL> with data as (select /*+ materialize */ distinct val as val from t where id = 1)
select max(id) from tmp tmp where tmp.sign = 0 and tmp.id in (select val from data) ;  2 
I thought about that. I’m reluctant to use undocumented hints such as materialize. So folowing query has almost the best plan for my data:
with data as (select distinct val as val from t where id = 1 and rownum > 0)
select
max(id)
from tmp tmp
where tmp.sign = 0
and tmp.id in (select * from data);
| Id  | Operation               | Name  | Rows  | Bytes | Cost (%CPU)| Time     |                                                                    
|   0 | SELECT STATEMENT        |       |     1 |     8 |    50   (8)| 00:00:01 |                                                                    
|   1 |  SORT AGGREGATE         |       |     1 |     8 |            |          |                                                                    
|   2 |   NESTED LOOPS          |       |     3 |    24 |    50   (8)| 00:00:01 |                                                                    
|   3 |    VIEW                 |       |     3 |     9 |    50   (8)| 00:00:01 |                                                                    
|   4 |     HASH UNIQUE         |       |     3 |    18 |    50   (8)| 00:00:01 |                                                                    
|   5 |      COUNT              |       |       |       |            |          |                                                                    
|*  6 |       FILTER            |       |       |       |            |          |                                                                    
|*  7 |        TABLE ACCESS FULL| T     | 50000 |   292K|    47   (3)| 00:00:01 |                                                                    
|*  8 |    INDEX RANGE SCAN     | TMP_I |     1 |     5 |     0   (0)| 00:00:01 |                                                                    
Predicate Information (identified by operation id):                                                                                                  
   6 - filter(ROWNUM>0)                                                                                                                              
   7 - filter("ID"=1)                                                                                                                                
   8 - access("TMP"."ID"="DATA"."VAL" AND "TMP"."SIGN"=0)   And changing two hidden parameters may lead to the same plan as I expect:
alter session set "_gby_hash_aggregation_enabled" = false;
alter session set "_simple_view_merging" = false;
with data as (select distinct val as val from t where id = 1)
select
max(id)
from tmp tmp
where tmp.sign = 0
and tmp.id in (select * from data);
| Id  | Operation             | Name  | Rows  | Bytes | Cost (%CPU)| Time     |                                                                      
|   0 | SELECT STATEMENT      |       |     1 |    18 |    50   (8)| 00:00:01 |                                                                      
|   1 |  SORT AGGREGATE       |       |     1 |    18 |            |          |                                                                      
|   2 |   NESTED LOOPS        |       |     3 |    54 |    50   (8)| 00:00:01 |                                                                      
|   3 |    VIEW               |       |     3 |    39 |    50   (8)| 00:00:01 |                                                                      
|   4 |     SORT UNIQUE       |       |     3 |    18 |    50   (8)| 00:00:01 |                                                                      
|*  5 |      TABLE ACCESS FULL| T     | 50000 |   292K|    47   (3)| 00:00:01 |                                                                      
|*  6 |    INDEX RANGE SCAN   | TMP_I |     1 |     5 |     0   (0)| 00:00:01 |                                                                      
Predicate Information (identified by operation id):                                                                                                  
   5 - filter("ID"=1)                                                                                                                                
   6 - access("TMP"."ID"="DATA"."VAL" AND "TMP"."SIGN"=0)   But here I've got two additional questions:
1. no_use_hash_aggregation can be used instead of "alter session set "_gby_hash_aggregation_enabled" = false;"
What hint can be used instead of "alter session set "_simple_view_merging" = false;"?
2. Is there any way to enforce CBO to use for this one
select
max(id)
from tmp tmp
where tmp.sign = 0
and tmp.id in (select distinct val as val from t where id = 1 and rownum > 0);
| Id  | Operation                   | Name  | Rows  | Bytes | Cost (%CPU)| Time     |                                                                
|   0 | SELECT STATEMENT            |       |     1 |     5 |     3   (0)| 00:00:01 |                                                                
|   1 |  SORT AGGREGATE             |       |     1 |     5 |            |          |                                                                
|   2 |   FIRST ROW                 |       |     1 |     5 |     1   (0)| 00:00:01 |                                                                
|*  3 |    INDEX FULL SCAN (MIN/MAX)| TMP_I |     1 |     5 |     1   (0)| 00:00:01 |                                                                
|*  4 |     FILTER                  |       |       |       |            |          |                                                                
|   5 |      COUNT                  |       |       |       |            |          |                                                                
|*  6 |       FILTER                |       |       |       |            |          |                                                                
|*  7 |        TABLE ACCESS FULL    | T     |     2 |    12 |     2   (0)| 00:00:01 |                                                                
-------------------------------------------------------------------------------------  the same plan as for this
with data as (select distinct val as val from t where id = 1 and rownum > 0)
select
max(id)
from tmp tmp
where tmp.sign = 0
and tmp.id in (select * from data);
| Id  | Operation               | Name  | Rows  | Bytes | Cost (%CPU)| Time     |                                                                    
|   0 | SELECT STATEMENT        |       |     1 |     8 |    50   (8)| 00:00:01 |                                                                    
|   1 |  SORT AGGREGATE         |       |     1 |     8 |            |          |                                                                    
|   2 |   NESTED LOOPS          |       |     3 |    24 |    50   (8)| 00:00:01 |                                                                    
|   3 |    VIEW                 |       |     3 |     9 |    50   (8)| 00:00:01 |                                                                    
|   4 |     HASH UNIQUE         |       |     3 |    18 |    50   (8)| 00:00:01 |                                                                    
|   5 |      COUNT              |       |       |       |            |          |                                                                    
|*  6 |       FILTER            |       |       |       |            |          |                                                                    
|*  7 |        TABLE ACCESS FULL| T     | 50000 |   292K|    47   (3)| 00:00:01 |                                                                    
|*  8 |    INDEX RANGE SCAN     | TMP_I |     1 |     5 |     0   (0)| 00:00:01 |                                                                    
I don't have anything against subquery factoring clause. Just for personal interest.
(I have read topic "Thread: Materialize a Subquery without using "with" clause"
Materialize a Subquery without using "with" clause
user503699 wrote:
If you are not going to change other things (like stats collection method, table/index structures etc.) which allow optimizer to choose better plan and you know your data better, you may need to be very specific with the hints and also may have to use additional hints in order to influence optimizer decisions.
One way to do that would be to get the base details from oracle as follows (and tweak them) :I didn't find keyword "ADVANCED" in specification for DISPLAY_CURSOR Function in documentation. Nice trick.
But anyway outline data makes sense only in case when query already has desirable execution plan.