查询优化不在左外连接中

Question

考虑以下表格;

测试

SQL> desc test;
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 NUM                                       NOT NULL NUMBER
 NUM2                                               NUMBER(10)
 NUM3                                               NUMBER


       NUM       NUM2       NUM3
---------- ---------- ----------
         1          1          1
         2          2          2

TEST2

SQL> desc test2;
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 NUM                                       NOT NULL NUMBER
 NUM2                                      NOT NULL NUMBER
 NUM3                                               NUMBER


       NUM       NUM2       NUM3
---------- ---------- ----------
         1          1          1
         3          1          1

根据这本书，如果要将NOT IN条件替换为LEFT外连接;查询性能会提高。

解释计划1

SQL> select * 
       from test 
      where num NOT IN (select num 
                          from test2);

NUM      NUM2       NUM3
--- --------- ----------
2          2          2

执行计划

----------------------------------------------------------
Plan hash value: 856752680

----------------------------------------------------------------------------
| Id  | Operation          | Name  | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |       |     1 |     2 |     5   (0)| 00:00:01 |
|*  1 |  FILTER            |       |       |       |            |          |
|   2 |   TABLE ACCESS FULL| TEST  |     2 |     4 |     3   (0)| 00:00:01 |
|*  3 |   TABLE ACCESS FULL| TEST2 |     2 |    26 |     2   (0)| 00:00:01 |
----------------------------------------------------------------------------

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

   1 - filter( NOT EXISTS (SELECT /*+ */ 0 FROM "TEST2" "TEST2" WHERE
              LNNVL("NUM"<>:B1)))
   3 - filter(LNNVL("NUM"<>:B1))

所以我在两个表的num列上创建了索引，并将查询重写为：

解释计划2

SQL> select * 
       from test 
  left join test2 on (test.num = test2.num) 
      where test2.num is null;

       NUM       NUM2       NUM3        NUM       NUM2       NUM3
---------- ---------- ---------- ---------- ---------- ----------
         2          2          2


Execution Plan
----------------------------------------------------------
Plan hash value: 1525288557

--------------------------------------------------------------------------------
| Id  | Operation                     | Name  | Rows  | Bytes | Cost (%CPU)| Tim
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT              |       |     2 |    82 |     4   (0)| 00:
|*  1 |  FILTER                       |       |       |       |            |
|   2 |   NESTED LOOPS OUTER          |       |     2 |    82 |     4   (0)| 00:
|   3 |    TABLE ACCESS FULL          | TEST  |     2 |     4 |     3   (0)| 00:
|   4 |    TABLE ACCESS BY INDEX ROWID| TEST2 |     1 |    39 |     1   (0)| 00:
|*  5 |     INDEX RANGE SCAN          | ID2   |     1 |       |     0   (0)| 00:
--------------------------------------------------------------------------------

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

   1 - filter("TEST2"."NUM" IS NULL)
   5 - access("TEST"."NUM"="TEST2"."NUM"(+))

我显然遗漏了一些东西，因为建议的方法证明更贵。这是由于我的数据分发吗？

Ver：Oracle 10g

Answer 1

您的两个查询返回不同的结果集。第二个版本也返回第二个表中的数据。我希望从两个表返回数据的查询比从一个表返回数据的查询更昂贵。

使用select test.*代替select *尝试第二个查询，看看它是否有所作为。