• 创建测试表

以用户jyu连接,创建测试表
SQL> conn jyu/jyu;
Connected. SQL> create table t (id number, name varchar2(100)); Table created. SQL> insert into t select rownum,object_name from dba_objects; 47391 rows created. SQL> commit; Commit complete. 创建索引
SQL> create index t_idx1 on t(id); Index created. 收集统计数据
SQL> exec dbms_stats.gather_table_stats('JYU','T'); PL/SQL procedure successfully completed.
  • 执行计划

查看SQL语句执行计划
SQL> set autotrace traceonly
SQL> select * from t where id=1; Execution Plan
----------------------------------------------------------
Plan hash value: 3292636276 --------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 28 | 2 (0)| 00:00:01 |
| 1 | TABLE ACCESS BY INDEX ROWID| T | 1 | 28 | 2 (0)| 00:00:01 |
|* 2 | INDEX RANGE SCAN | T_IDX1 | 1 | | 1 (0)| 00:00:01 |
-------------------------------------------------------------------------------------- Predicate Information (identified by operation id):
--------------------------------------------------- 2 - access("ID"=1) Note
-----
- outline "OLD_OUTLN" used for this statement Statistics
----------------------------------------------------------
0 recursive calls
0 db block gets
4 consistent gets
0 physical reads
0 redo size
576 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
1 rows processed
SQL语句选择了使用索引的执行计划 使用Hint指定语句使用全表扫描的执行计划
SQL> explain plan for select /*+ full(t) */ * from t where id=1; Explained. SQL> select * from table(dbms_xplan.display); PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------
Plan hash value: 2153619298 --------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 28 | 50 (2)| 00:00:01 |
|* 1 | TABLE ACCESS FULL| T | 1 | 28 | 50 (2)| 00:00:01 |
-------------------------------------------------------------------------- Predicate Information (identified by operation id):
--------------------------------------------------- PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------ 1 - filter("ID"=1) 13 rows selected.

使用outline固定执行计划

以sysdba连接数据库
SQL> conn /as sysdba
Connected. 分别为2个SQL语句创建outline
SQL> alter session set current_schema = jyu; Session altered. SQL> create or replace outline OLD_OUTLN for category TEMP_PLAN on select * from t where id=1; Outline created. SQL> create or replace outline NEW_OUTLN for category TEMP_PLAN on select /*+ full(t) */ * from t where id=1; Outline created. 交换SQL语句的outline
SQL> create private outline OLFROM from OLD_OUTLN; Outline created. SQL> create private outline OLTO from NEW_OUTLN; Outline created. SQL> update ol$ set hintcount=(select hintcount from ol$ where ol_name='OLTO') where ol_name='OLFROM'; 1 row updated. SQL> delete from ol$ where ol_name='OLTO'; 1 row deleted. SQL> update ol$ set ol_name='OLTO' where ol_name='OLFROM'; 1 row updated. SQL> commit; Commit complete. SQL> execute dbms_outln_edit.refresh_private_outline('OLTO'); PL/SQL procedure successfully completed. SQL> create or replace outline OLD_OUTLN from private OLTO for category GOOD_PLAN; Outline created. SQL> drop outline NEW_OUTLN; Outline dropped.

  • 设置使用指定的outlines

有两种方式可在全局设置使用outline

方式一:使用alter system设置(数据库重启后失效)
SQL> conn / as sysdba
Connected.
SQL> alter system set use_stored_outlines=GOOD_PLAN; System altered. 方式二:通过trigger设置(数据库重启仍然有效)
SQL> create or replace trigger enable_outlines_trig
--Ref : How to Enable USE_STORED_OUTLINES Permanently (Doc ID 560331.1)
after startup on database
begin
execute immediate('alter system set use_stored_outlines=GOOD_PLAN');
end;
/
  • 检查SQL语句执行计划

SQL> conn jyu/jyu
Connected.
SQL> set autotrace traceonly
SQL> select * from t where id=1; Execution Plan
----------------------------------------------------------
Plan hash value: 2153619298 --------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 28 | 50 (2)| 00:00:01 |
|* 1 | TABLE ACCESS FULL| T | 1 | 28 | 50 (2)| 00:00:01 |
-------------------------------------------------------------------------- Predicate Information (identified by operation id):
--------------------------------------------------- 1 - filter("ID"=1) Note
-----
- outline "OLD_OUTLN" used for this statement Statistics
----------------------------------------------------------
34 recursive calls
145 db block gets
269 consistent gets
0 physical reads
576 redo size
576 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
2 sorts (memory)
0 sorts (disk)
1 rows processed

outline生效,SQL语句执行了全表扫描。

#!/bin/bash

if [ $# -lt 3 ]; then
cat <<EOF
Fix given SQL plan using given sql in file
usage: fix_plan.sh <hash_value> <hinted_sql_file> <owner>
examples: fix_plan.sh 716428968 good.sql ALEX
EOF
exit 1
fi HASH_VALUE=$1
SQL_FILE=$2
OL_OWNER=$3 echo "HASH_VALUE : $HASH_VALUE"
echo "SQL_FILE : $SQL_FILE"
echo "OL_OWNER : $OL_OWNER"
echo "" #Create outline by hash_value
function create_ol_from_hashvalue {
HASH_VALUE=$1
OL_NAME=$2
OL_OWNER=$3 #generate create outline sql
#I didn't use dbms_outln.create_outline, because it cannot create given name outline
# and there's no hash value in V$SQL and DBA_OUTLINES to associate the two
# according to "How To Match a SQL Statement to a Stored Outline (Doc ID 743312.1)"
sqlplus -S "/ as sysdba" > /tmp/tmp_$OL_NAME.sql <<EOF
set feedback off
set serveroutput on size unlimited
declare
v_sqltext varchar2(32000);
begin
--get sql text
select dbms_lob.substr(SQL_FULLTEXT, 30000, 1 ) into v_sqltext from v\$sql where hash_value = $HASH_VALUE and rownum=1; dbms_output.put_line('alter session set current_schema = $OL_OWNER;');
v_sqltext := 'create or replace outline $OL_NAME for category TEMP_PLAN on ' || chr(10) || v_sqltext || chr(10) ||';';
dbms_output.put_line(v_sqltext);
dbms_output.put_line('exit;');
end;
/
EOF sqlplus -S "/ as sysdba" @/tmp/tmp_$OL_NAME.sql
} #Create outline from sql file
function create_ol_from_sqlfile {
SQL_FILE=$1
OL_NAME=$2
OL_OWNER=$3 #generate create outline sql
cat > /tmp/tmp_$OL_NAME.sql <<EOF
alter session set current_schema = $OL_OWNER;
create or replace outline $OL_NAME for category TEMP_PLAN on
`cat $SQL_FILE`
exit;
EOF sqlplus -S "/ as sysdba" @/tmp/tmp_$OL_NAME.sql } #Exchange outlines, make GOOD_SQL plan to GOOD_PLAN category
#Ref: How to Edit a Stored Outline to Use the Plan from Another Stored Outline (Doc ID 730062.1)
function exchange_outline {
OL1=$1
OL2=$2
OL_OWNER=$3 sqlplus -S "/ as sysdba" <<EOF
set feedback off
alter session set current_schema = $OL_OWNER;
create private outline OLFROM from $OL1;
create private outline OLTO from $OL2;
update ol$ set hintcount=(select hintcount from ol$ where ol_name='OLTO') where ol_name='OLFROM';
delete from ol$ where ol_name='OLTO';
update ol$ set ol_name='OLTO' where ol_name='OLFROM';
commit;
execute dbms_outln_edit.refresh_private_outline('OLTO');
create or replace outline $OL1 from private OLTO for category GOOD_PLAN;
drop outline $OL2;
exit;
EOF } #display outline
function display_outline {
OL_NAME=$1
OL_OWNER=$2 sqlplus -S "/ as sysdba" <<EOF
set pagesize 1000 linesize 160
set long 32000
col hint format a55
col join_pos format a45
col owner format a12
col name format a18
col ts format a14
col h format 999
col category format a12
col sql_text format a80
col used format a6 select name, sql_text, category, used, to_char(TIMESTAMP, 'YY-mm-dd hh24:MI')
from dba_outlines
where name = '$OL_NAME' and OWNER = '$OL_OWNER'; select ol_name name, category, hint#, stage# stage, hint_text hint, join_pred join_pos
from outln.ol\$hints
where ol_name = '$OL_NAME'
order by ol_name, hint#;
exit;
EOF
} #main function
echo "1. Create outline OL_$HASH_VALUE for SQL $HASH_VALUE"
create_ol_from_hashvalue $HASH_VALUE OL_$HASH_VALUE $OL_OWNER echo "2. Create outline OL_TEMP for SQL in $SQL_FILE"
create_ol_from_sqlfile $SQL_FILE OL_TEMP $OL_OWNER echo "3. Exchange outline OL_$HASH_VALUE with OL_TEMP, and drop OL_TEMP"
exchange_outline OL_$HASH_VALUE OL_TEMP $OL_OWNER echo "4. Display final outline for SQL $HASH_VALUE : OL_$HASH_VALUE in category GOOD_PLAN "
display_outline OL_$HASH_VALUE $OL_OWNER

如何通过outline为SQL语句指定执行计划的更多相关文章

  1. oracle中查看sql语句的执行计划

    1.在pl/sql中打开cmd命令容器 2.在cmd命令窗口中输入:explain plan for select * from t; 3.查看sql语句的执行计划:select * from tab ...

  2. 【Oracle】三种方式查看SQL语句的执行计划

    查看执行计划的方式有三种: EXPLAIN PLAN .V$SQL_PLAN .SQL*PLUS AUTOTRACE 1.EXPLAIN PLAN: 显示执行相应语句时可以使用的理论计划 读取执行计划 ...

  3. SQL语句的执行计划(oracle表的三种链接方式)

    SQL语句我们写完之后,就是分析其优化,这就要求我们了解到底数据是怎么存储. 首先我们需要了解,表链接的几种方式 nested loop join sort merge join hash join ...

  4. 【测试】使用hr用户下的employees表写一条SQL语句,执行计划走索引全扫描

    SQL> select count(*) from employees; COUNT(*) ---------- Execution Plan ------------------------- ...

  5. 【测试】并使用scott用户下的emp表写一条SQL语句,执行计划走唯一索引

    SQL; SAL ---------- Execution Plan ---------------------------------------------------------- ------ ...

  6. ORACLE数据库SQL语句的执行过程

    SQL语句在数据库中处理过程是怎样的呢?执行顺序呢?在回答这个问题前,我们先来回顾一下:在ORACLE数据库系统架构下,SQL语句由用户进程产生,然后传到相对应的服务端进程,之后由服务器进程执行该SQ ...

  7. 官方文档:11G新特性SQL PLAN BASLINE 执行计划基线

    什么是SQL执行计划管理? SQL计划管理(SQL plan management)是一咱预防机制,记录和评估SQL语句的执行计划.SQL plan management的主要功能是sql plan ...

  8. 详细分析SQL语句逻辑执行过程和相关语法

    本文目录: 1.SQL语句的逻辑处理顺序 1.2 各数据库系统的语句逻辑处理顺序 1.2.1 SQL Server和Oracle的逻辑执行顺序 1.2.2 MariaDB的逻辑执行顺序 1.2.3 M ...

  9. mybatis下使用log4j打印sql语句和执行结果

    转载自:https://www.cnblogs.com/jeevan/p/3493972.html 本来以为很简单的问题, 结果自己搞了半天还是不行; 然后google, baidu, 搜出来各种方法 ...

随机推荐

  1. 扩展KMP(转)

    拓展kmp是对KMP算法的扩展,它解决如下问题: 定义母串S,和字串T,设S的长度为n,T的长度为m,求T与S的每一个后缀的最长公共前缀,也就是说,设extend数组,extend[i]表示T与S[i ...

  2. 解析远程域名主机的IP地址

    我们知道,计算机在访问远程主机的时候,本质上是通过IP地址来进行访问的,但我们实际在使用的时候,例如我们想访问百度的主页,我们是通过在浏览器的地址栏输入百度的域名来进行访问的,因此,计算机需要将百度的 ...

  3. JavaScript -- 操作input CheckBox 全选框

    <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/ ...

  4. QT 中文乱码问题

    1. 在main函数中创建完 QApplication对象后马上添加 QTextCodec::setCodecForTr(QTextCodec::codecForName("UTF-8&qu ...

  5. linux-RabbitMQ安装命令

    一.RabbitMQ 1.安装配置epel源    $ rpm -ivh http://dl.fedoraproject.org/pub/epel/6/i386/epel-release-6-8.no ...

  6. OpenStack with Opendaylight Part 1: Intro to Pipeline

    Using Vagrant to create vm nodes; devstack to start openstack using Opendaylight as ML2. Openstack w ...

  7. tensorflow笔记:多层CNN代码分析

    tensorflow笔记系列: (一) tensorflow笔记:流程,概念和简单代码注释 (二) tensorflow笔记:多层CNN代码分析 (三) tensorflow笔记:多层LSTM代码分析 ...

  8. 微信开放接口获取用户昵称保存到MySQL中为空白

    微信昵称中包含emoji表情标签,某些标签是使用了4字节编码的UTF8. 而大多数MySQL数据库现在使用的是3字节UTF8编码,这样会导致保存为空,且不会提示失败. 解决方法有2个,一个是升级到My ...

  9. 《Think in Java》(九)接口

    接口和内部类为我们提供了一种将接口与实现分离的更加结构化的方法: 抽象化类则是普通类与接口之间的一种中庸之道: 涨姿势了 接口也可以拥有值属性,但它们都是隐式的 static 和 final 的: 接 ...

  10. Java面试题下

    这部分主要是开源Java EE框架方面的内容,包括hibernate.MyBatis.spring.Spring MVC等,由于Struts 2已经是明日黄花,在这里就不讨论Struts 2的面试题, ...