Oracle里的执行计划
Posted 春困秋乏夏打盹
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了Oracle里的执行计划相关的知识，希望对你有一定的参考价值。
1 执行计划的查看
1 explan plan
2 dbms_xplan包
3 sqlplus中 set autot
4 10046
5 awr报告
6一些现成的脚本
7 10053
常见的就前4种
执行计划的补充信息
Note
-----
- dynamic sampling used for this statement
- cardinality feedback used for this statement
- sql profile sys_sqlprof_ used for this statement
1 explain plan
Plsql developer中F5,Toad中ctrl+e，
create table testyhq (id1 number,name1 varchar2(20),date1 date default sysdate); create table testyhq2 (id1 number,name1 varchar2(20),date1 date default sysdate); insert into testyhq (id1,name1) values (1,\'yhq\'); insert into testyhq2 (id1,name1) values (1,\'yhq\'); explain plan for select * from testyhq t1 where exists (select 1 from testyhq2 t2 where t1.id1=t2.id1); select * from table(dbms_xplan.display);
2 dbms_xplan包
select * from table(dbms_xplan.display) select * from table(dbms_xplan.display_cursor(null,null,\'runstats_last\'));-all,advanced select * from table(dbms_xplan.display_cursor(null,null,\'allstats last\')); select * from table(dbms_xplan.display_cursor(\'dwcqthpyuzbgm\',null,\'typical -predicate -rows\'));
3 autot on
set autot [off|on traceonly] [exp] [stat]
4 10046跟tkprof命令
--详细见10046
2 如何得到真实的执行计划
判断执行计划是否真实，要根据真正执行过
select count(*) from emp where ename=\'hongquan\';
select * from v$sqlarea where sql_text like\'select count(*) from emp%\'
--得到 sql_id
select * from table(dbms_xplan.display_cursor(\'dg5qqawqhkbay\',null,\'typical -predicate -rows\'));---advanced
-得到真实的执行计划
- Warning: basic plan statistics not available. These are only collected when:
* hint \'gather_plan_statistics\' is used for the statement or
* parameter \'statistics_level\' is set to \'ALL\', at session or system level
SQL> conn scott Enter password: Connected. SQL> set linesize 1000 SQL> set pagesize 1000 SQL> SELECT ename,dname,loc 2 FROM emp e, dept d 3 WHERE e.deptno = d.deptno 4 AND e.empno = 7788; ENAME DNAME LOC ---------- -------------- ------------- SCOTT RESEARCH DALLAS SQL> select * from table(dbms_xplan.display_cursor(null,null)); PLAN_TABLE_OUTPUT SQL_ID bbms3v8t5j4zr, child number 0 ------------------------------------- SELECT ename,dname,loc FROM emp e, dept d WHERE e.deptno = d.deptno AND e.empno = 7788 Plan hash value: 2385808155 ---------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ---------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | | | 2 (100)| | | 1 | NESTED LOOPS | | 1 | 33 | 2 (0)| 00:00:01 | | 2 | TABLE ACCESS BY INDEX ROWID| EMP | 1 | 13 | 1 (0)| 00:00:01 | |* 3 | INDEX UNIQUE SCAN | PK_EMP | 1 | | 0 (0)| | | 4 | TABLE ACCESS BY INDEX ROWID| DEPT | 4 | 80 | 1 (0)| 00:00:01 | |* 5 | INDEX UNIQUE SCAN | PK_DEPT | 1 | | 0 (0)| | ---------------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 3 - access("E"."EMPNO"=7788) 5 - access("E"."DEPTNO"="D"."DEPTNO") 24 rows selected. SQL> select sql_id,address,plan_hash_value,hash_value,child_number from v$sql 2 where sql_text like \'%SELECT ename%\' and sql_text not like \'%from v$sql%\'; SQL_ID ADDRESS PLAN_HASH_VALUE HASH_VALUE CHILD_NUMBER ------------- -------- --------------- ---------- ------------ d3kymurnb00vk 309E4398 2385808155 3903849330 0 bbms3v8t5j4zr 3094A978 2385808155 844665847 0 SQL> select * from table(dbms_xplan.display_cursor(\'bbms3v8t5j4zr\',null,\'typical -predicate -rows\')); PLAN_TABLE_OUTPUT SQL_ID bbms3v8t5j4zr, child number 0 ------------------------------------- SELECT ename,dname,loc FROM emp e, dept d WHERE e.deptno = d.deptno AND e.empno = 7788 Plan hash value: 2385808155 -------------------------------------------------------------------------------- | Id | Operation | Name | Bytes | Cost (%CPU)| Time | -------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | | 2 (100)| | | 1 | NESTED LOOPS | | 33 | 2 (0)| 00:00:01 | | 2 | TABLE ACCESS BY INDEX ROWID| EMP | 13 | 1 (0)| 00:00:01 | | 3 | INDEX UNIQUE SCAN | PK_EMP | | 0 (0)| | | 4 | TABLE ACCESS BY INDEX ROWID| DEPT | 80 | 1 (0)| 00:00:01 | | 5 | INDEX UNIQUE SCAN | PK_DEPT | | 0 (0)| | -------------------------------------------------------------------------------- 18 rows selected. SQL> show parameter statistics_le NAME TYPE VALUE ------------------------------------ ----------- ------------------------------ statistics_level string TYPICAL SQL> alter session set statistics_level=all; Session altered. SQL> select e.ename,e.sal,s.grade 2 from emp e 3 join salgrade s 4 on e.sal between losal and hisal 5 and e.deptno = 20; ENAME SAL GRADE ---------- ---------- ---------- SCOTT 3000 4 FORD 3000 4 JONES 2975 4 ADAMS 1100 1 SMITH 800 1 SQL> select * from table(dbms_xplan.display_cursor(null,null,\'iostats last -predicate -note\')); PLAN_TABLE_OUTPUT SQL_ID d75njv99u8m2n, child number 0 ------------------------------------- select e.ename,e.sal,s.grade from emp e join salgrade s on e.sal between losal and hisal and e.deptno = 20 Plan hash value: 4204027666 ---------------------------------------------------------------------------------------------------- | Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers | Reads | -------------------------------------------------------------------------------------------- | 1 | MERGE JOIN | | 1 | 1 | 5 |00:00:00.04 | 14 | 11 | | 2 | SORT JOIN | | 1 | 5 | 5 |00:00:00.03 | 7 | 5 | | 3 | TABLE ACCESS FULL | EMP | 1 | 5 | 5 |00:00:00.03 | 7 | 5 | | 4 | FILTER | | 5 | | 5 |00:00:00.01 | 7 | 6 | | 5 | SORT JOIN | | 5 | 5 | 14 |00:00:00.01 | 7 | 6 | | 6 | TABLE ACCESS FULL| SALGRADE | 1 | 5 | 5 |00:00:00.01 | 7 | 6 | ---------------------------------------------------------------------------------------------------- 18 rows selected. SQL> alter session set statistics_level=typical; Session altered. SQL> SELECT /*+ gather_plan_statistics */ ename,dname,loc 2 FROM emp e, dept d 3 WHERE e.deptno = d.deptno 4 AND d.deptno=20 ORDER BY 1,2,3; ENAME DNAME LOC ---------- -------------- ------------- ADAMS RESEARCH DALLAS FORD RESEARCH DALLAS JONES RESEARCH DALLAS SCOTT RESEARCH DALLAS SMITH RESEARCH DALLAS SQL> select * from table(dbms_xplan.display_cursor(null,null,\'allstats -rows\')); PLAN_TABLE_OUTPUT SQL_ID cdazxuhca6j88, child number 0 ------------------------------------- SELECT /*+ gather_plan_statistics */ ename,dname,loc FROM emp e, dept d WHERE e.deptno = d.deptno AND d.deptno=20 ORDER BY 1,2,3 Plan hash value: 3339094711 | Id | Operation | Name | Starts | A-Rows | A-Time | Buffers | OMem | 1Mem | O/1/M | --------------------------------------------------------------------------------------------------------------------- | 1 | SORT ORDER BY | | 1 | 5 |00:00:00.01 | 9 | 2048 | 2048 | 1/0/0| | 2 | NESTED LOOPS | | 1 | 5 |00:00:00.01 | 9 | | | | | 3 | TABLE ACCESS BY INDEX ROWID| DEPT | 1 | 1 |00:00:00.01 | 2 | | | | |* 4 | INDEX UNIQUE SCAN | PK_DEPT | 1 | 1 |00:00:00.01 | 1 | | | | |* 5 | TABLE ACCESS FULL | EMP | 1 | 5 |00:00:00.01 | 7 | | | | -------------------------------------------------------------------------------------------- Predicate Information (identified by operation id): --------------------------------------------------- 4 - access("D"."DEPTNO"=20) 5 - filter("E"."DEPTNO"=20) 23 rows selected. SQL>
3 查看执行计划的执行顺序
同级，最上，最右，
---- ----安装一个xplan
select * from table(xplan.display_cursor(null,null,\'runstats_last\'));
4 oracle里常见的执行计划
1 与表访问相关
TABLE ACCESS FULL
TABLE ACCESS BY USER ROWID
TABLE ACCESS BY INDEX ROWID
2与b tree index 相关
INDEX UNIQURE INDEX SCAN
INDEX RANGE SCAN
INDEX FULL SCAN
INDEX FAST FULL SCAN
INDEX SKIP SCAN
3与表连接相关
SORT JOIN ,MERGE JOIN
NESTED LOOPS,
HASH JOIN,
NESTED LOOPS ANTI,SEMI,MERGE JOIN
Anti：not in，not exists
Semi：in，exists----半连接会对连接结果进行必要的去重
--分区
--http://www.dbaxiaoyu.com/archives/1780
与位图索引相关的执行计划
位图索引主要用于数据仓库或者dss系统，位图索引实现了快捷的按位运算
位图索引没有行锁的概念，要锁就锁索引行的整个位图段，多个数据行可能对应于同一个索引行的位图段
位图索引：如果索引行的distinct值比较少，那么与b树索引比较，会占用较少的存储空间，能够快速处理and，or的条件的sql
位图索引，会存储为null值的列（及时是单键值）
5 其他典型的执行计划
1 and-equal(index merge)
Where出了多个不同单列的等值条件，并且多列上有单键的index，则会根据相应的单个index去扫描这些index
SQL> set linesize 800 SQL> set pagesize 1000 SQL> set autot trace SQL> create index idx_empno on emp_t(empno); Index created. SQL> create index idx_mgr on emp_t(mgr); Index created. SQL> select /*+ and_equal( emp_t idx_empno idx_mgr)*/ empno,job,ename,mgr from emp_t where empno=7566 and mgr=7839; Execution Plan ---------------------------------------------------------- Plan hash value: 3378047620 ----------------------------------------------------------------------------------------- | Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | ----------------------------------------------------------------------------------------- | 0 | SELECT STATEMENT | | 1 | 22 | 2 (0)| 00:00:01 | |* 1 | TABLE ACCESS BY INDEX ROWID| EMP_T | 1 | 22 | 2 (0)| 00:00:01 | | 2 | AND-EQUAL | | | | | | |* 3 | INDEX RANGE SCAN | IDX_EMPNO | 1 | | 1 (0)| 00:00:01 | |* 4 | INDEX RANGE SCAN | IDX_MGR | 2 | | 1 (0)| 00:00:Oracle执行计划里的 access和filter有什么区别