Oracle里的常见的执行计划

首页 > 代码库 > Oracle里的常见的执行计划

Oracle里的常见的执行计划

2024-08-10 04:55:52 224人阅读

与表访问相关的执行计划
Oracle数据库里面与表访问有关的的两种方法：全表扫描和ROWID扫描。反应在执行计划上，与全表扫描对应的执行计划中的关键字是“TABLE ACCESS FULL”,与ROWID扫描对应的执行计划中的关键字是"TABLE ACCESS BY USER ROWID"或“TABLE ACCESS BY INDEX ROWID”.
我们来看一下与表访问的相关的执行计划，先执行如下SQL:

SQL> select empno,ename,rowid from emp where ename=‘TURNER‘;

EMPNO ENAME ROWID

---------- ---------- ------------------

7844 TURNER AAAVREAAEAAAACXAAJ

SQL> select * from table(dbms_xplan.display_cursor(null,null,‘ALL‘));

PLAN_TABLE_OUTPUT

--------------------------------------------------------------------------------

SQL_ID 3bjd8ps607cau, child number 0

-------------------------------------

select empno,ename,rowid from emp where ename=‘TURNER‘

Plan hash value: 3956160932

--------------------------------------------------------------------------

--------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | | | 3 (100)| |

|* 1 | TABLE ACCESS FULL| EMP | 1 | 22 | 3 (0)| 00:00:01 |

从上述显示内容中可以看出，目标sql的执行计划走的是对表EMP的全表扫描，全表扫描在执行计划中对应的关键字就是“TABLE ACCESS FULL”

将上述sql改写成以指定的ROWID的方式执行：

SQL> select empno,ename from emp where rowid=‘AAAVREAAEAAAACXAAJ‘;

EMPNO ENAME

---------- ----------

7844 TURNER

SQL> select * from table(dbms_xplan.display_cursor(null,null,‘ALL‘));

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

SQL_ID 8n08pmh16ud05, child number 0

-------------------------------------

select empno,ename from emp where rowid=‘AAAVREAAEAAAACXAAJ‘

Plan hash value: 1116584662

-----------------------------------------------------------------------------------

-----------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | | | 1 (100)| |

| 1 | TABLE ACCESS BY USER ROWID| EMP | 1 | 22 | 1 (0)| 00:00:01 |

从上述显示内容可以看出，现在该sql的执行计划走的对表emp的rowid扫描，其对应的执行计划关键字“TABLE ACCESS BY USER ROWID”。

注意，ROWID扫描所对应的的执行计划的关键字还有可能是“TABLE ACCESS BY INDEX ROWID”,这取决于访问表时的ROWID来源。如果ROWID是来源于用户手工指定（例如上述指定“select empno,ename from emp where rowid=‘AAAVREAAEAAAACXAAJ‘”），则对应的执行计划关键字是“TABLE ACCESS BY USER ROWID”；如果是ROWID是来源于索引，则对应的执行计划关键字是“TABLE ACCESS BY INDEX ROWID”

表EMP的主键是列EMPNO（即列EMPNO上有主键索引），我们将目标sql改写成如下形式后执行：

SQL> select empno,ename from emp where empno=7369;

EMPNO ENAME

---------- ----------

7369 SMITH

SQL> select * from table(dbms_xplan.display_cursor(null,null,‘ALL‘));

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

SQL_ID 6yzqcfbz5xz3c, child number 0

-------------------------------------

select empno,ename from emp where empno=7369

Plan hash value: 2949544139

--------------------------------------------------------------------------------------

--------------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | | | 1 (100)| |

| 1 | TABLE ACCESS BY INDEX ROWID| EMP | 1 | 10 | 1 (0)| 00:00:01 |

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

|* 2 | INDEX UNIQUE SCAN | PK_EMP | 1 | | 0 (0)| |

--------------------------------------------------------------------------------------

从上述显示内容可以看出，此时目标sql的执行计划在访问表emp走的是对表emp的rowid扫描，因为这里的rowid是来源于索引PK_emp,所以其对应的执行计划关键字是“TABLE ACCESS BY INDEX ROWID”.

2.与B树索引相关的执行计划

常见的与B树索引访问相关的方法有：包括索引唯一扫描，索引范围扫描，索引全扫描，索引快速全扫描和索引跳跃式扫描，反应在执行计划上，与索引唯一扫描对应的关键字“INDEX UNIQUE SCAN”,与索引范围扫描对应的关键字是“INDEX RANGE SCAN”,与索引全扫描对应的关键字是“INDEX FULL SCAN”,与索引快速全扫描对应的关键字是“INDEX FAST FULL SCAN”,与索引跳跃式扫描对应的关键字是“INDEX SKIP SCAN”.

下面来看一下与B树索引访问相关的的执行计划实例。创建一个测试表EMPLOYEE:

SQL> create table employee(gender varchar2(1),employee_id number);

Table created.

SQL> insert into employee values(‘F‘,‘99‘);

1 row created.

SQL> insert into employee values(‘F‘,‘100‘);

1 row created.

SQL> insert into employee values(‘M‘,‘101‘);

1 row created.

SQL> insert into employee values(‘M‘,‘102‘);

1 row created.

SQL> insert into employee values(‘M‘,‘103‘);

1 row created.

SQL> insert into employee values(‘M‘,‘104‘);

1 row created.

SQL> insert into employee values(‘M‘,‘105‘);

1 row created.

SQL> commit;

Commit complete.

SQL> create unique index idx_uni_emp on employee(employee_id);

Index created.

SQL> select * from employee where employee_id=100;

G EMPLOYEE_ID

- -----------

F 100

SQL> select plan_table_output from table(dbms_xplan.display_cursor(null,null,‘ALL‘));

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

SQL_ID bum8qv24s6tqp, child number 0

-------------------------------------

select * from employee where employee_id=100

Plan hash value: 1887894887

-------------------------------------------------------------------------------------------

-------------------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | | | 1 (100)| |

| 1 | TABLE ACCESS BY INDEX ROWID| EMPLOYEE | 1 | 15 | 1 (0)| 00:00:01 |

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

|* 2 | INDEX UNIQUE SCAN | IDX_UNI_EMP | 1 | | 0 (0)| |

-------------------------------------------------------------------------------------------

从上述显示内容可以看出，此sql的执行计划走的是索引IDX_UNI_EMP的索引唯一扫描，索引唯一扫描在执行计划中对应的关键字就是“INDEX UNIQUE SCAN”.

Drop掉上述唯一索引IDX_UNI_EMP

SQL> drop index idx_uni_emp;

Index dropped.

SQL> create index idx_emp_1 on employee(employee_id);

Index created.

SQL> select * from employee where employee_id=100;

G EMPLOYEE_ID

- -----------

F 100

SQL> select plan_table_output from table(dbms_xplan.display_cursor(null,null,‘ALL‘));

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

SQL_ID bum8qv24s6tqp, child number 0

-------------------------------------

select * from employee where employee_id=100

Plan hash value: 2428325319

-----------------------------------------------------------------------------------------

-----------------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | | | 2 (100)| |

| 1 | TABLE ACCESS BY INDEX ROWID| EMPLOYEE | 1 | 15 | 2 (0)| 00:00:01 |

PLAN_TABLE_OUTPUT

-------------------------------------------------------------------------------------------

|* 2 | INDEX RANGE SCAN | IDX_EMP_1 | 1 | | 1 (0)| 00:00:01 |

-----------------------------------------------------------------------------------------

从上述显示内容中可以看出，现在改sql的执行计划走是对索引idx_emp_1的索引范围扫描，索引范围扫描在执行计划中对应的关键字就是“INDEX RANGE SCAN”。

truncate表EMPLOYEE中的数据：

SQL> truncate table employee;

Table truncated.

更新插入10000条记录：

SQL> begin

2 for i in 1..5000 loop

3 insert into employee values(‘F‘,i);

4 end loop;

5 commit;

6 end;

7 /

PL/SQL procedure successfully completed.

SQL> begin

2 for i in 5001..10000 loop

3 insert into employee values(‘M‘,i);

4 end loop;

5 commit;

6 end;

7 /

PL/SQL procedure successfully completed.

SQL> select gender,count(*) from employee group by gender;

G COUNT(*)

- ----------

M 5000

F 5000

对表EMPLOYEE收集一下统计信息：

SQL> exec dbms_stats.gather_table_stats(ownname=>‘SCOTT‘,tabname=>‘EMPLOYEE‘,estimate_percent=>100,cascade=>TRUE,no_invalidate=>false,method_opt=>‘FOR ALL COLUMNS SIZE 1‘);

PL/SQL procedure successfully completed.

SQL> set autotrace traceonly

SQL> show user

USER is "SCOTT"

SQL> select employee_id from employee;

10000 rows selected.

Execution Plan

----------------------------------------------------------

Plan hash value: 2119105728

------------------------------------------------------------------------------

------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 10000 | 40000 | 7 (0)| 00:00:01 |

| 1 | TABLE ACCESS FULL| EMPLOYEE | 10000 | 40000 | 7 (0)| 00:00:01 |

------------------------------------------------------------------------------

Statistics

----------------------------------------------------------

1 recursive calls

0 db block gets

689 consistent gets

0 physical reads

0 redo size

174308 bytes sent via SQL*Net to client

7850 bytes received via SQL*Net from client

668 SQL*Net roundtrips to/from client

0 sorts (memory)

0 sorts (disk)

10000 rows processed

明明上述SQL查询字段employee_id可以通过扫描索引idx_emp_1得到，但oracle依然选择了对employee的全表扫描。

此时就算我们使用Hint强制让oracle扫描索引idx_emp_1，从如下结果可以看到，oracle依然选择了对表employee的全表扫描（即hint失效了）

SQL> select /* index(employee idx_emp_1)*/employee_id from employee;

10000 rows selected.

Execution Plan

----------------------------------------------------------

Plan hash value: 2119105728

------------------------------------------------------------------------------

------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 10000 | 40000 | 7 (0)| 00:00:01 |

| 1 | TABLE ACCESS FULL| EMPLOYEE | 10000 | 40000 | 7 (0)| 00:00:01 |

------------------------------------------------------------------------------

Statistics

----------------------------------------------------------

1 recursive calls

0 db block gets

689 consistent gets

0 physical reads

0 redo size

174308 bytes sent via SQL*Net to client

7850 bytes received via SQL*Net from client

668 SQL*Net roundtrips to/from client

0 sorts (memory)

0 sorts (disk)

10000 rows processed

会出现上述现象是因为oracle无论如何总会保证目标sql结果的正确性，可能会得到错误结果的执行路径orale是不会考虑的。对于idx_emp_1而言，它是一个单键值的B树索引，索引NULL值不会存储在其中，那么一旦列employee_id中出现null值（虽然这里实际上并没有null值），则扫描索引IDX_EMP_1的结果就会漏掉那些employee_id为NULL的值，这也就意味这个如果orale在执行上述sql选择了扫描索引idx_emp_1，那么执行结果就可能是不准的。这种情况下，oracle当然不会考虑扫描idx_emp_1，即使我们使用Hint。

如果这里我们想让oracle在执行上述sql扫描索引idx_emp_1，则必须将列employee_id的属性改成not null。这就相当于告诉oracle，这里列employee_id上不会有null值，你就放心的扫描idx_emp_1吧。

SQL> select employee_id from employee;

10000 rows selected.

Execution Plan

----------------------------------------------------------

Plan hash value: 3918702848

----------------------------------------------------------------------------------

----------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 10000 | 40000 | 7 (0)| 00:00:01 |

| 1 | INDEX FAST FULL SCAN| IDX_EMP_1 | 10000 | 40000 | 7 (0)| 00:00:01 |

----------------------------------------------------------------------------------

从上述显示内容可以看出，现在sql的执行计划走的是idx_emp_1的索引快速全扫描，索引快速全扫描在执行计划中对应的关键字就是“INDEX FAST FULL SCAN”

现在我们加上强制索引IDX_EMP_1的hint，再次执行该sql：

SQL> select /*+ index(employee idx_emp_1) */ employee_id from employee;

10000 rows selected.

Execution Plan

----------------------------------------------------------

Plan hash value: 438557521

------------------------------------------------------------------------------

------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 10000 | 40000 | 20 (0)| 00:00:01 |

| 1 | INDEX FULL SCAN | IDX_EMP_1 | 10000 | 40000 | 20 (0)| 00:00:01 |

------------------------------------------------------------------------------

从上述显示内容可以看出，现在SQL的执行计划走的是对索引idx_emp_1的索引全扫描，索引全扫描在执行计划中对应的关键字就是“INDEX FULL SCAN”.

DROP掉单键值B树索引IDX_EMP_1；

SQL> drop index idx_emp_1;

Index dropped.

SQL> create index index_emp_2 on employee(gender,employee_id);

Index created.

SQL> set autot trace

SQL> select * from employee where employee_id=101;

Execution Plan

----------------------------------------------------------

Plan hash value: 2052968723

--------------------------------------------------------------------------------

--------------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 1 | 6 | 3 (0)| 00:00:01 |

|* 1 | INDEX SKIP SCAN | INDEX_EMP_2 | 1 | 6 | 3 (0)| 00:00:01 |

--------------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

1 - access("EMPLOYEE_ID"=101)

filter("EMPLOYEE_ID"=101)

Statistics

----------------------------------------------------------

0 recursive calls

0 db block gets

6 consistent gets

0 physical reads

0 redo size

600 bytes sent via SQL*Net to client

524 bytes received via SQL*Net from client

2 SQL*Net roundtrips to/from client

0 sorts (memory)

0 sorts (disk)

1 rows processed

从上述显示内容可以看出，现在sql的执行计划走的是对索引IDX_EMP_2的索引跳跃式扫描，索引跳跃式扫描在执行计划中对应的关键字就是“INDEX SKIP SCAN”.

Oracle里的常见的执行计划

声明：以上内容来自用户投稿及互联网公开渠道收集整理发布，本网站不拥有所有权，未作人工编辑处理，也不承担相关法律责任，若内容有误或涉及侵权可进行投诉：投诉/举报工作人员会在5个工作日内联系你，一经查实，本站将立刻删除涉嫌侵权内容。

联系
我们

首页 > 代码库 > Oracle里的常见的执行计划

Oracle里的常见的执行计划

看完仍有疑问？有类似问题直接问程序猿