使用SQL Profile及SQL Tuning Advisor固定执行计划
SQL Profiles可以说是Outlines的进化。Outlines能够实现的功能SQL Profiles也完全能够实现,而SQL Profiles具有Outlines不具备的优化,最重要的有二点:
SQL Profiles更容易生成、更改和控制。
SQL Profiles在对SQL语句的支持上做得更好,也就是适用范围更广。
使用SQL Profiles两个目的:
锁定或者说是稳定执行计划。
在不能修改应用中的SQL的情况下使SQL语句按指定的执行计划运行。
参考老熊的博客,链接:http://www.laoxiong.net/sql-profiles-partii.html http://www.laoxiong.net/sql-profiles-part.html http://blog.sina.com.cn/s/blog_5037eacb01011mgu.htm
1.实验环境构造语句:
create table bys.t1 as select * from dba_objects;create index bys.t1_idx on t1(object_id);
exec dbms_stats.gather_table_stats(‘bys‘,‘t1‘,cascade=>true,degree=>4);
set autotrace trace;
select a.*,b.owner from t1 a,t1 b where a.object_name like ‘%T1%‘ and a.object_id=b.object_id;
select /*+ use_nl(a b) index(t2) */ a.*,b.owner from t1 a,t1 b where a.object_name like ‘%T1%‘ and a.object_id=b.object_id;
set autotrace off;
explain plan for select /*+ use_nl(a b) index(t2) */ a.*,b.owner from t1 a,t1 b where a.object_name like ‘%T1%‘ and a.object_id=b.object_id;
col sql_text for a100
SELECT SQL_ID,SQL_TEXT FROM V$SQL WHERE SQL_TEXT LIKE ‘%a.*,b.owner from t1 a,t1 b%‘;
####2t5xqt4d1dsaw
10g开始,v$sql_plan中就包括了SQL语句OUTLINE数据,也就是稳定执行计划的Hints。如下:
set pagesize 1000
SQL> select * from table(dbms_xplan.display_cursor(‘2t5xqt4d1dsaw‘,null,‘outline‘));
PLAN_TABLE_OUTPUT
---------------------------------------------------------------------------------------------------------------
SQL_ID 2t5xqt4d1dsaw, child number 0
-------------------------------------
select /*+ use_nl(a b) index(t2) */ a.*,b.owner from t1 a,t1 b where
a.object_name like ‘%T1%‘ and a.object_id=b.object_id
Plan hash value: 190596302
--------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | | | 979 (100)| |
| 1 | TABLE ACCESS BY INDEX ROWID| T1 | 1 | 9 | 2 (0)| 00:00:01 |
| 2 | NESTED LOOPS | | 471 | 43803 | 979 (1)| 00:00:12 |
|* 3 | TABLE ACCESS FULL | T1 | 471 | 39564 | 36 (0)| 00:00:01 |
|* 4 | INDEX RANGE SCAN | T1_IDX | 1 | | 1 (0)| 00:00:01 |
--------------------------------------------------------------------------------------
Outline Data ----------这一部分: /*+ */ 之间的就是所需的数据了
-------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE(‘10.2.0.1‘)
ALL_ROWS
OUTLINE_LEAF(@"SEL$1")
FULL(@"SEL$1" "A"@"SEL$1")
INDEX(@"SEL$1" "B"@"SEL$1" ("T1"."OBJECT_ID"))
LEADING(@"SEL$1" "A"@"SEL$1" "B"@"SEL$1")
USE_NL(@"SEL$1" "B"@"SEL$1")
END_OUTLINE_DATA
*/
PLAN_TABLE_OUTPUT
-----------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - filter("A"."OBJECT_NAME" LIKE ‘%T1%‘)
4 - access("A"."OBJECT_ID"="B"."OBJECT_ID")
###############################################################################
2.开始创建并应用SQL Profile
使用未加HINT的SQL语句,并指定SQL_PROFILE名字:SQLPROFILE_T1注意事项:两个引号---如(‘10.2.0.1‘)要写成(‘‘10.2.0.1‘‘)--‘%T1%‘写成‘‘%T1%‘‘
SQL语句结尾的;不要写上,绑定变量值是‘1‘时,要输入‘‘1‘‘
declare
v_hints sys.sqlprof_attr;
begin
v_hints:=sys.sqlprof_attr(
‘BEGIN_OUTLINE_DATA‘,
‘IGNORE_OPTIM_EMBEDDED_HINTS‘,
‘OPTIMIZER_FEATURES_ENABLE(‘‘10.2.0.1‘‘)‘,
‘ALL_ROWS‘,
‘OUTLINE_LEAF(@"SEL$1")‘,
‘FULL(@"SEL$1" "A"@"SEL$1")‘,
‘INDEX(@"SEL$1" "B"@"SEL$1" ("T1"."OBJECT_ID"))‘,
‘LEADING(@"SEL$1" "A"@"SEL$1" "B"@"SEL$1")‘,
‘USE_NL(@"SEL$1" "B"@"SEL$1")‘,
‘END_OUTLINE_DATA‘);
dbms_sqltune.import_sql_profile(
‘select a.*,b.owner from t1 a,t1 b where a.object_name like ‘‘%T1%‘‘ and a.object_id=b.object_id‘,
v_hints,‘SQLPROFILE_T1‘,
force_match=>true,replace=>true);
end;
/
删除语句:
exec dbms_sqltune.drop_sql_profile(name =>‘SQLPROFILE_T1‘ );
######
3.验证SQL语句是否使用上一步创建的SQL_PROFILE:
SQL> set autotrace trace;SQL> select a.*,b.owner from t1 a,t1 b where a.object_name like ‘%T1%‘ and a.object_id=b.object_id;
23 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 190596302
--------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 471 | 43803 | 979 (1)| 00:00:12 |
| 1 | TABLE ACCESS BY INDEX ROWID| T1 | 1 | 9 | 2 (0)| 00:00:01 |
| 2 | NESTED LOOPS | | 471 | 43803 | 979 (1)| 00:00:12 |
|* 3 | TABLE ACCESS FULL | T1 | 471 | 39564 | 36 (0)| 00:00:01 |
|* 4 | INDEX RANGE SCAN | T1_IDX | 1 | | 1 (0)| 00:00:01 |
--------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - filter("A"."OBJECT_NAME" LIKE ‘%T1%‘)
4 - access("A"."OBJECT_ID"="B"."OBJECT_ID")
Note
-----
- SQL profile "SQLPROFILE_T1" used for this statement
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
163 consistent gets
0 physical reads
0 redo size
2609 bytes sent via SQL*Net to client
392 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
23 rows processed 0 sorts (disk)
#####################################################################
使用Oracle的SQL Tuning Advisor固定执行计划--SQL_ID
----通过sql_id的方式 sql_id => ‘a2h6pzvqncfvg‘,--想调优的sql_id
set serveroutput on;
DECLARE
a_tuning_task VARCHAR2(30);
BEGIN
a_tuning_task := dbms_sqltune.create_tuning_task(sql_id => ‘a2h6pzvqncfvg‘,task_name => ‘tuning_test‘);
dbms_sqltune.execute_tuning_task(a_tuning_task);
END;
/
set long 1000
SELECT dbms_sqltune.report_tuning_task(‘tuning_test‘) FROM dual;
这里的输出中的sql_id是从v$sql查出,对应的是没有加hint的SQL。
如果SQL Tuning Advisor找到了理想的执行计划,下一步就是:Accept SQL Profile,接受这个SQL Profile。
## execute dbms_sqltune.accept_sql_profile(task_name =>‘tuning_test‘ ,replace => TRUE,force_match=>true);
完成后验证语句:
set autotrace trace;
select a.*,b.owner from t1 a,t1 b where a.object_name like ‘%T1%‘ and a.object_id=b.object_id;
set autotrace off;
删除SQL语句:
exec dbms_sqltune.drop_tuning_task(‘tuning_test‘);
如果未找到理想的执行计划,则参考下一个方法。
##################################################################
使用Oracle的SQL Tuning Advisor固定执行计划--使用SQL_TEXT,参考以下:
DECLAREmy_task_name VARCHAR2 (30);
my_sqltext CLOB;
BEGIN
my_sqltext := ‘select * from test where OBJECT_ID=15‘; --想调优的sql
my_task_name := dbms_sqltune.create_tuning_task (sql_text=> my_sqltext,
--bind_list => ‘UNDO$‘,
--user_name => ‘SONG‘,
--scope => ‘COMPREHENSIVE‘,
--time_limit => 60,
task_name => ‘sql_tuning_test‘
--description => ‘Tuning Task‘
);
END;
/
exec dbms_sqltune.execute_tuning_task(‘sql_tuning_test‘); --执行调优的任务
SELECT STATUS FROM USER_ADVISOR_TASKS WHERE task_name = ‘sql_tuning_test‘; --查看调优作业的状态
SELECT DBMS_SQLTUNE.REPORT_TUNING_TASK(‘sql_tuning_test‘) FROM DUAL;--查看调优的建议
exec dbms_sqltune.drop_tuning_task(‘sql_tuning_test‘);--删除任务
郑重声明:本站内容如果来自互联网及其他传播媒体,其版权均属原媒体及文章作者所有。转载目的在于传递更多信息及用于网络分享,并不代表本站赞同其观点和对其真实性负责,也不构成任何其他建议。