not in与not exists性能比较

基础语法：

not in的写法中，内外表都要全表扫描
select count * from test where object_name not in (select object_name from test2);

not exists的写法中，内层表子查询可以走索引
select count * from test2 t2 where not exists(select 1 from test t1 where t1.object_name=t2.object_name );
（object_name上建有有索引）

执行计划：

再来看一下CBO模式下的情况：
SQL> select * from test2
  2  where table_name not in
  3        ( select object_name from test1 where object_name like 'A%') ;  
2950 rows selected.
Elapsed: 00:00:13.93

Execution Plan
----------------------------------------------------------
Plan hash value: 833525739
-------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |                 |  1576 |   315K|  6364   (1)| 00:01:17 |
|*  1 |  FILTER            |                 |       |       |            |     |
|   2 |   TABLE ACCESS FULL| TEST2           |  1577 |   315K|    14   (0)| 00:00:01 |
|*  3 |   INDEX RANGE SCAN | IDX_OBJECT_NAME |     1 |    25 |     8   (0)| 00:00:01 |
---------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - filter( NOT EXISTS (SELECT /*+ */ 0 FROM "TEST1" "TEST1" WHERE
              "OBJECT_NAME" LIKE 'A%' AND LNNVL("OBJECT_NAME"<>:B1)))
   3 - access("OBJECT_NAME" LIKE 'A%')
       filter("OBJECT_NAME" LIKE 'A%' AND LNNVL("OBJECT_NAME"<>:B1))

Statistics
----------------------------------------------------------
          1  recursive calls
          0  db block gets
      60806  consistent gets
         29  physical reads
       1104  redo size
     165661  bytes sent via SQL*Net to client
       2541  bytes received via SQL*Net from client
        198  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
       2950  rows processed

SQL> select * from test2
  2  where not exists
  3        ( select 1 from test1 where test1.object_name = test2.table_name and test1.object_name like 'A%') ;
2950 rows selected.
Elapsed: 00:00:00.19

Execution Plan
----------------------------------------------------------
Plan hash value: 349475409

----------------------------------------------------------------------------------------
| Id  | Operation            | Name            | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |                 |  1538 |   345K|    23   (5)| 00:00:01 |
|*  1 |  HASH JOIN RIGHT ANTI|                 |  1538 |   345K|    23   (5)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN   | IDX_OBJECT_NAME |  1259 | 31475 |     8   (0)| 00:00:01 |
|   3 |   TABLE ACCESS FULL  | TEST2           |  1577 |   315K|    14   (0)| 00:00:01 |
----------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
   1 - access("TEST1"."OBJECT_NAME"="TEST2"."TABLE_NAME")
   2 - access("TEST1"."OBJECT_NAME" LIKE 'A%')
       filter("TEST1"."OBJECT_NAME" LIKE 'A%')

Statistics
----------------------------------------------------------
          1  recursive calls
          0  db block gets
        314  consistent gets
          0  physical reads
          0  redo size
     165661  bytes sent via SQL*Net to client
       2541  bytes received via SQL*Net from client
        198  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
       2950  rows processed

在CBO模式下，not in采用的是filter，对外层表做全表扫描，再去filter内层查询结果集。not exists采用内层查询结果集作hash join连接。
从结果明显可以看到，not exists效率比较高。把内层和外层的表对换一下，结果也是相同的，not exists的效率明显高很多。