Oracle 使用 sql profile 固定执行计划

测试使用 sql profile 固定执行计划:

Oracle 10g之前有outlines,10g之后 sql profile 。如果针对非绑定变量的sql,outlines则效果不佳,不建议使用 。

1、准备测试用表

SQL> create table zzh_ob as select * from dba_objects;

SQL> create index ind_obid on zzh_ob(object_id);

SQL> select object_id from zzh_ob where rownum<2;

OBJECT_ID


16

SQL> exec dbms_stats.gather_table_stats(user,'zzh_ob',cascade=>true);

原sql执行计划:

SQL> set autot trace explain

SQL> select * from zzh_ob where object_id=16;

Execution Plan


Plan hash value: 3485916696


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


| 0 | SELECT STATEMENT | | 1 | 127 | 2 (0)| 00:00:01 |

| 1 | TABLE ACCESS BY INDEX ROWID BATCHED| ZZH_OB | 1 | 127 | 2 (0)| 00:00:01 |

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


Predicate Information (identified by operation id):


2 - access("OBJECT_ID"=16)

新sql执行计划

SQL> select /*+ full(zzh_ob) */ * from zzh_ob where object_id=16;

Execution Plan


Plan hash value: 474274488


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


| 0 | SELECT STATEMENT | | 1 | 127 | 388 (1)| 00:00:01 |

|* 1 | TABLE ACCESS FULL| ZZH_OB | 1 | 127 | 388 (1)| 00:00:01 |


Predicate Information (identified by operation id):


1 - filter("OBJECT_ID"=16)

2、获取新sql的sql_id

SQL> col sql_id for a20

SQL> col sql_text for a120

SQL> select sql_id,sql_text from v$sql where sql_text like '%full(zzh_ob)%';

SQL_ID SQL_TEXT


g5btxvsh51ct5 EXPLAIN PLAN SET STATEMENT_ID='PLUS1200001' FOR select sql_id,sql_text from v$sql where sql_text like '%full(zzh_ob)%'

4kn9w9q3xf73k EXPLAIN PLAN SET STATEMENT_ID='PLUS1200001' FOR select /*+ full(zzh_ob) */ * from zzh_ob where object_id=16

68x7hd7uaqqk1 select sql_id,sql_text from v$sql where sql_text like '%full(zzh_ob)%'

1syfgv2q5ggn2 select /*+ full(zzh_ob) */ * from zzh_ob where object_id=16

3、获取新sql的outline

SQL> set pagesize 1000

SQL> select * from table(dbms_xplan.display_cursor('1syfgv2q5ggn2',null,'outline'));

PLAN_TABLE_OUTPUT


SQL_ID 1syfgv2q5ggn2, child number 0


select /*+ full(zzh_ob) */ * from zzh_ob where object_id=16

Plan hash value: 474274488


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


| 0 | SELECT STATEMENT | | | | 388 (100)| |

|* 1 | TABLE ACCESS FULL| ZZH_OB | 1 | 127 | 388 (1)| 00:00:01 |

PLAN_TABLE_OUTPUT



Outline Data


/*+

BEGIN_OUTLINE_DATA

IGNORE_OPTIM_EMBEDDED_HINTS

OPTIMIZER_FEATURES_ENABLE('19.1.0')

DB_VERSION('19.1.0')

ALL_ROWS

PLAN_TABLE_OUTPUT


OUTLINE_LEAF(@"SEL$1")

FULL(@"SEL1" ["ZZH_OB"@"SEL1](mailto:)")

END_OUTLINE_DATA

*/

Predicate Information (identified by operation id):


1 - filter("OBJECT_ID"=16)

32 rows selected.

4、创建sql profile(SQLPROFILE_01)

SQL> declare

v_hints sys.sqlprof_attr;

begin

v_hints:=sys.sqlprof_attr(

'BEGIN_OUTLINE_DATA',

'IGNORE_OPTIM_EMBEDDED_HINTS',

'OPTIMIZER_FEATURES_ENABLE(''19.1.0'')',

'DB_VERSION(''19.1.0'')',

'ALL_ROWS',

'OUTLINE_LEAF(@"SEL$1")',

'FULL(@"SEL1" ["ZZH_OB"@"SEL1")'](mailto:),

'END_OUTLINE_DATA');

dbms_sqltune.import_sql_profile(

'select * from zzh_ob where object_id=16',

v_hints,'SQLPROFILE_01',

force_match=>true,replace=>false);

end;

/

5、查看是否使用 sql profile

SQL> set autot trace explain

SQL> select * from zzh_ob where object_id=16;

Execution Plan


Plan hash value: 474274488


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


| 0 | SELECT STATEMENT | | 1 | 127 | 388 (1)| 00:00:01 |

|* 1 | TABLE ACCESS FULL| ZZH_OB | 1 | 127 | 388 (1)| 00:00:01 |


Predicate Information (identified by operation id):


1 - filter("OBJECT_ID"=16)

Note


  • SQL profile "SQLPROFILE_01" used for this statement

Statistics


247 recursive calls

0 db block gets

1544 consistent gets

2 physical reads

0 redo size

2685 bytes sent via SQL*Net to client

401 bytes received via SQL*Net from client

2 SQL*Net roundtrips to/from client

11 sorts (memory)

0 sorts (disk)

1 rows processed

查询已经创建的 sql profile :

SELECT name, created, category, sql_Text from dba_sql_profiles ORDER BY created DESC;

相关推荐
NineData1 小时前
数据库迁移总踩坑?用 NineData 迁移评估,提前识别所有兼容性风险
数据库·程序员·云计算
阿里云大数据AI技术1 小时前
用 SQL 调大模型?Hologres + 百炼,让数据开发直接“对话”AI
sql·llm
赵渝强老师3 小时前
【赵渝强老师】PostgreSQL中表的碎片
数据库·postgresql
全栈老石7 小时前
拆解低代码引擎核心:元数据驱动的"万能表"架构
数据库·低代码
倔强的石头_1 天前
kingbase备份与恢复实战(二)—— sys_dump库级逻辑备份与恢复(Windows详细步骤)
数据库
jiayou642 天前
KingbaseES 实战:深度解析数据库对象访问权限管理
数据库
李广坤3 天前
MySQL 大表字段变更实践(改名 + 改类型 + 改长度)
数据库
爱可生开源社区4 天前
2026 年,优秀的 DBA 需要具备哪些素质?
数据库·人工智能·dba
随逸1774 天前
《从零搭建NestJS项目》
数据库·typescript
加号35 天前
windows系统下mysql多源数据库同步部署
数据库·windows·mysql