[20180808]exists and not exists.txt

[20180808]exists and not exists.txt

--//生产系统遇到的一个性能问题,通过例子来说明:

1.环境:
SCOTT@test01p> @ ver1

PORT_STRING                    VERSION        BANNER                                                                               CON_ID
------------------------------ -------------- -------------------------------------------------------------------------------- ----------
IBMPC/WIN_NT64-9.1.0           12.1.0.1.0     Oracle Database 12c Enterprise Edition Release 12.1.0.1.0 - 64bit Production              0

SCOTT@test01p> create table t1 as select * from all_objects;
Table created.

SCOTT@test01p> create table t2 as select object_id,'1' flag from t1;
Table created.

SCOTT@test01p> select max(object_id) from t2;
MAX(OBJECT_ID)
--------------
        107828

SCOTT@test01p> update t2 set flag='0' where object_id=107828;
1 row updated.

SCOTT@test01p> commit ;
Commit complete.

SCOTT@test01p> create index i_t2_flag on t2(flag);
Index created.

--//分析表,并且t2的flag字段建立直方图.
execute sys.dbms_stats.gather_table_stats ( OwnName => user,TabName => 't1',Estimate_Percent => NULL,Method_Opt => 'FOR ALL COLUMNS SIZE 1 ',Cascade => True ,No_Invalidate => false);
execute sys.dbms_stats.gather_table_stats ( OwnName => user,TabName => 't2',Estimate_Percent => NULL,Method_Opt => 'FOR ALL COLUMNS SIZE 1 for columns flag size 10  ',Cascade => True ,No_Invalidate => false);
`
2.测试:
SCOTT@test01p> alter session set statistics_level=all;
Session altered.

SCOTT@test01p> select object_name  from t1 where not exists (select 1 from t2 where t2.object_id=t1.object_id and t2.flag='1' );
OBJECT_NAME
--------------------
T1

SCOTT@test01p> @ dpc '' ''
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID  d4qcxhmwy49r1, child number 0
-------------------------------------
select object_name  from t1 where not exists (select 1 from t2 where
t2.object_id=t1.object_id and t2.flag='1' )

Plan hash value: 629543484

-------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id  | Operation            | Name | Starts | E-Rows |E-Bytes|E-Temp | Cost (%CPU)| E-Time   | A-Rows |   A-Time   | Buffers | Reads  |  OMem |  1Mem | Used-Mem |
-------------------------------------------------------------------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |      |      1 |        |       |       |   728 (100)|          |      1 |00:00:00.24 |    1667 |   1511 |       |       |          |
|*  1 |  HASH JOIN RIGHT ANTI|      |      1 |    899 | 33263 |  1672K|   728   (1)| 00:00:01 |      1 |00:00:00.24 |    1667 |   1511 |  5536K|  3056K| 5658K (0)|
|*  2 |   TABLE ACCESS FULL  | T2   |      1 |  89876 |   614K|       |    46   (3)| 00:00:01 |  89876 |00:00:00.02 |     152 |      0 |       |       |          |
|   3 |   TABLE ACCESS FULL  | T1   |      1 |  89877 |  2633K|       |   421   (1)| 00:00:01 |  89877 |00:00:00.11 |    1515 |   1511 |       |       |          |
-------------------------------------------------------------------------------------------------------------------------------------------------------------------

Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------

1 - SEL$5DA710D3
   2 - SEL$5DA710D3 / T2@SEL$2
   3 - SEL$5DA710D3 / T1@SEL$1

Predicate Information (identified by operation id):
---------------------------------------------------

1 - access("T2"."OBJECT_ID"="T1"."OBJECT_ID")
   2 - filter("T2"."FLAG"='1')

--//仔细看id-2.过滤条件是   2 - filter("T2"."FLAG"='1').这样即使你建立索引在t2.flag也不会使用.因为flag='1'占大多数.
--//实际上对于当前应用改成如下是等效的.因为flag仅仅两种取值'0','1'.

SCOTT@test01p> select object_name  from t1 where  exists (select 1 from t2 where t2.object_id=t1.object_id and t2.flag='0' );
OBJECT_NAME
--------------------
T1

SCOTT@test01p> @ dpc '' ''
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID  1y5xvtwz0u11f, child number 0
-------------------------------------
select object_name  from t1 where  exists (select 1 from t2 where
t2.object_id=t1.object_id and t2.flag='0' )
Plan hash value: 1273788863
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id  | Operation                            | Name      | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time   | A-Rows |   A-Time   | Buffers | Reads  |  OMem |  1Mem | Used-Mem |
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT                     |           |      1 |        |       |   423 (100)|          |      1 |00:00:00.19 |    1518 |   1512 |       |       |          |
|*  1 |  HASH JOIN RIGHT SEMI                |           |      1 |      1 |    37 |   423   (1)| 00:00:01 |      1 |00:00:00.19 |    1518 |   1512 |  2168K|  2168K|  697K (0)|
|   2 |   TABLE ACCESS BY INDEX ROWID BATCHED| T2        |      1 |      1 |     7 |     2   (0)| 00:00:01 |      1 |00:00:00.04 |       3 |      1 |       |       |          |
|*  3 |    INDEX RANGE SCAN                  | I_T2_FLAG |      1 |      1 |       |     1   (0)| 00:00:01 |      1 |00:00:00.04 |       2 |      1 |       |       |          |
|   4 |   TABLE ACCESS FULL                  | T1        |      1 |  89877 |  2633K|   421   (1)| 00:00:01 |  89877 |00:00:00.12 |    1515 |   1511 |       |       |          |
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
   1 - SEL$5DA710D3
   2 - SEL$5DA710D3 / T2@SEL$2
   3 - SEL$5DA710D3 / T2@SEL$2
   4 - SEL$5DA710D3 / T1@SEL$1
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - access("T2"."OBJECT_ID"="T1"."OBJECT_ID")
   3 - access("T2"."FLAG"='0')

--//实际上到具体应用object_id字段是主键,如果在上面建立索引,逻辑读更小.
CREATE UNIQUE INDEX SCOTT.pk_t1 ON SCOTT.T1 (OBJECT_ID);
ALTER TABLE SCOTT.T1 ADD CONSTRAINT pk_t1  PRIMARY KEY (OBJECT_ID);

CREATE UNIQUE INDEX SCOTT.pk_t2 ON SCOTT.T2 (OBJECT_ID);
ALTER TABLE SCOTT.T2 ADD CONSTRAINT pk_t2  PRIMARY KEY (OBJECT_ID);

SCOTT@test01p> select object_name  from t1 where  exists (select 1 from t2 where t2.object_id=t1.object_id and t2.flag='0' );
OBJECT_NAME
--------------------
T1

SCOTT@test01p> @ dpc '' ''
PLAN_TABLE_OUTPUT
-------------------------------------
SQL_ID  1y5xvtwz0u11f, child number 0
-------------------------------------
select object_name  from t1 where  exists (select 1 from t2 where
t2.object_id=t1.object_id and t2.flag='0' )
Plan hash value: 4193600567
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Id  | Operation                              | Name      | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time   | A-Rows |   A-Time   | Buffers | Reads  |  OMem |  1Mem | Used-Mem |
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT                       |           |      1 |        |       |     3 (100)|          |      1 |00:00:00.04 |       6 |      2 |       |       |          |
|   1 |  NESTED LOOPS                          |           |      1 |        |       |            |          |      1 |00:00:00.04 |       6 |      2 |       |       |          |
|   2 |   NESTED LOOPS                         |           |      1 |      1 |    37 |     3   (0)| 00:00:01 |      1 |00:00:00.03 |       5 |      1 |       |       |          |
|   3 |    SORT UNIQUE                         |           |      1 |      1 |     7 |     2   (0)| 00:00:01 |      1 |00:00:00.01 |       3 |      0 |  2048 |  2048 | 2048  (0)|
|   4 |     TABLE ACCESS BY INDEX ROWID BATCHED| T2        |      1 |      1 |     7 |     2   (0)| 00:00:01 |      1 |00:00:00.01 |       3 |      0 |       |       |          |
|*  5 |      INDEX RANGE SCAN                  | I_T2_FLAG |      1 |      1 |       |     1   (0)| 00:00:01 |      1 |00:00:00.01 |       2 |      0 |       |       |          |
|*  6 |    INDEX UNIQUE SCAN                   | PK_T1     |      1 |      1 |       |     0   (0)|          |      1 |00:00:00.03 |       2 |      1 |       |       |          |
|   7 |   TABLE ACCESS BY INDEX ROWID          | T1        |      1 |      1 |    30 |     1   (0)| 00:00:01 |      1 |00:00:00.01 |       1 |      1 |       |       |          |
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
   4 - SEL$5DA710D3 / T2@SEL$2
   5 - SEL$5DA710D3 / T2@SEL$2
   6 - SEL$5DA710D3 / T1@SEL$1
   7 - SEL$5DA710D3 / T1@SEL$1
Predicate Information (identified by operation id):
---------------------------------------------------
   5 - access("T2"."FLAG"='0')
   6 - access("T2"."OBJECT_ID"="T1"."OBJECT_ID")
Note
-----
   - this is an adaptive plan

--//而select object_name  from t1 where not exists (select 1 from t2 where t2.object_id=t1.object_id and t2.flag='1' );执行计划不变.不再贴出.
--//我有时候想开发写sql代码过脑子没有,有时候真的很无语很无奈...