用 SQL Tuning Advisor (STA) 优化SQL语句

      转自http://blog.csdn.net/tianlesoftware/article/details/5630888

在Oracle10g之前，优化SQL是个比较费力的技术活，不停的分析执行计划，加hint，分析统计信息等等。在10g中，Oracle推出了自己的SQL优化辅助工具: SQL优化器(SQL Tuning Advisor ：STA)，它是新的DBMS_SQLTUNE包。使用STA一定要保证优化器是CBO模式下。

执行DBMS_SQLTUNE包进行sql优化需要有advisor的权限：

SQL> create table bigtab as select rownum as "id",a.* from sys.all_objects a;

SQL>  create table smalltab as select rownum as "id", a.* FROM sys.all_tables a;

SQL> set autot on

SQL> select count(1) from bigtab a,smalltab b where a.object_name=b.table_name;

  COUNT(1)

----------

      1675

执行计划

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

Plan hash value: 3089226980

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

| Id  | Operation           | Name     | Rows  | Bytes | Cost (%CPU)| Time     |

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

|   0 | SELECT STATEMENT    |          |     1 |    34 |   178   (3)| 00:00:03 |

|   1 |  SORT AGGREGATE     |          |     1 |    34 |            |          |

|*  2 |   HASH JOIN         |          | 15405 |   511K|   178   (3)| 00:00:03 |

|   3 |    TABLE ACCESS FULL| SMALLTAB |  1535 | 26095 |    13   (0)| 00:00:01 |

|   4 |    TABLE ACCESS FULL| BIGTAB   | 48623 |   807K|   163   (2)| 00:00:02 |

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

Predicate Information (identified by operation id):

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

   2 - access("A"."OBJECT_NAME"="B"."TABLE_NAME")

Note------ dynamic sampling used for this statement

统计信息

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

          9  recursive calls

          0  db block gets

        881  consistent gets

          0  physical reads

          0  redo size

        409  bytes sent via SQL*Net to client

        385  bytes received via SQL*Net from client

          2  SQL*Net roundtrips to/from client

          2  sorts (memory)

          0  sorts (disk)

          1  rows processed

SQL> set autot off

SQL> set timing off

SQL> DECLARE
  2       my_task_name VARCHAR2(30);

  3        my_sqltext   CLOB;

  4      BEGIN

  5        my_sqltext := 'select count(*) from bigtab a, smalltab b where a.object_name=b.table_name';

  6       my_task_name := DBMS_SQLTUNE.CREATE_TUNING_TASK(

  7              sql_text    => my_sqltext,

  8              user_name   => 'TEST',   -- 注意是大写，不然会报错，用户无效

  9               scope       => 'COMPREHENSIVE',

 10             time_limit  => 60,

 11             task_name   => 'tuning_sql_test',

 12             description => 'Task to tune a query on a specified table');

 13

 14     DBMS_SQLTUNE.EXECUTE_TUNING_TASK( task_name => 'tuning_sql_test');

 15   END;

 16    /

PL/SQL 过程已成功完成。

SQL> SELECT task_name,status FROM USER_ADVISOR_TASKS WHERE task_name ='tuning_sql_test';

TASK_NAME                      STATUS

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

tuning_sql_test                COMPLETED

SQL> SET LONG 999999

SQL> set serveroutput on size 999999

SQL> SET LINESIZE 100

SQL> SELECT DBMS_SQLTUNE.REPORT_TUNING_TASK( 'tuning_sql_test') from DUAL;

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

GENERAL INFORMATION SECTION

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

Tuning Task Name                  : tuning_sql_test

Tuning Task Owner                 : TEST

Scope                             : COMPREHENSIVE

Time Limit(seconds)               : 60

Completion Status                 : COMPLETED

Started at                        : 08/22/2012 20:24:44

Completed at                      : 08/22/2012 20:24:45

Number of Statistic Findings      : 2

Number of Index Findings          : 1

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

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

Schema Name: TEST

SQL ID     : 6p64dnnsqf9pm

SQL Text   : select count(*) from bigtab a, smalltab b where

             a.object_name=b.table_name

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

FINDINGS SECTION (3 findings)

-------------------------------------------------------------------------------
DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

--------------------------------------------------------------------------------
1- Statistics Finding

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

尚未分析表 "TEST"."SMALLTA

  Recommendation

  --------------

  - 考虑收集此表的优

    execute dbms_stats.gather_table_stats(ownname => 'TEST', tabname =>

            'SMALLTAB', estimate_percent => DBMS_STATS.AUTO_SAMPLE_SIZE,

            method_opt => 'FOR ALL COLUMNS SIZE AUTO');

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

  Rationale

  ---------

为了选择好的执行计划, 优化程序需

2- Statistics Finding

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

尚未分析表 "TEST"."BIGTA

  Recommendation

  --------------

  - 考虑收集此表的优

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

    execute dbms_stats.gather_table_stats(ownname => 'TEST', tabname =>

            'BIGTAB', estimate_percent => DBMS_STATS.AUTO_SAMPLE_SIZE,

            method_opt => 'FOR ALL COLUMNS SIZE AUTO');

  Rationale

  ---------

为了选择好的执行计划, 优化程序需

3- Index Finding (see explain plans section below)

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

通过创建一个或多个索引可以

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

  Recommendation (estimated benefit: 100%)

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

  -考虑运行可以改进物理方案设计的 Access Advi

    create index TEST.IDX$$_00330001 on TEST.SMALLTAB('TABLE_NAME');

  -考虑运行可以改进物理方案设计的 Access Advi

    create index TEST.IDX$$_00330002 on TEST.BIGTAB('OBJECT_NAME');

  Rationale

  ---------

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

创建推荐的索引可以显著地改进此语句的执行计划。但是, 使用典型的

 可能比单个语句更可取。通过这种方法可以获得全面的索引建

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

EXPLAIN PLANS SECTION

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

1- Original

-----------

Plan hash value: 3089226980

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

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

| Id  | Operation           | Name     | Rows  | Bytes | Cost (%CPU)| Time     |

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

|   0 | SELECT STATEMENT    |          |     1 |    34 |   178   (3)| 00:00:03 |

|   1 |  SORT AGGREGATE     |          |     1 |    34 |            |          |

|*  2 |   HASH JOIN         |          | 15405 |   511K|   178   (3)| 00:00:03 |

|   3 |    TABLE ACCESS FULL| SMALLTAB |  1535 | 26095 |    13   (0)| 00:00:01 |

|   4 |    TABLE ACCESS FULL| BIGTAB   | 48623 |   807K|   163   (2)| 00:00:02 |

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

Predicate Information (identified by operation id):

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

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

   2 - access("A"."OBJECT_NAME"="B"."TABLE_NAME")

2- Using New Indices

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

Plan hash value: 105601825

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

| Id  | Operation              | Name           | Rows  | Bytes | Cost (%CPU)| Time

 |

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

|   0 | SELECT STATEMENT       |                |     1 |    34 |    62   (5)| 00:00:01 |

|   1 |  SORT AGGREGATE        |                |     1 |    34 |            |          |

|*  2 |   HASH JOIN            |                | 15405 |   511K|    62   (5)| 00:00:01

 |

|   3 |    INDEX FAST FULL SCAN| IDX$$_00330001 |  1535 | 26095 |     6   (0)| 00:00:01 |

|   4 |    INDEX FAST FULL SCAN| IDX$$_00330002 | 48623 |   807K|    55   (4)| 00:00:01 |

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

Predicate Information (identified by operation id):

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

DBMS_SQLTUNE.REPORT_TUNING_TASK('TUNING_SQL_TEST')

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

   2 - access("A"."OBJECT_NAME"="B"."TABLE_NAME")

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

通过调用dbms_sqltuen.drop_tuning_task可以删除已经存在的优化任务

SQL>exec dbms_sqltune.drop_tuning_task('tuning_sql_test');

PL/SQL procedure successfully completed.