讀懂執(zhí)行計(jì)劃

查看執(zhí)行計(jì)劃的方法

1. Explain Plan For SQL
   ? 不實(shí)際執(zhí)行SQL語(yǔ)句，生成的計(jì)劃未必是真實(shí)執(zhí)行的計(jì)劃
   ? 必須要有plan_table
   是一種手段但不是最好的
2. SQLPLUS AUTOTRACE
   ? 除set autotrace traceonly explain外均實(shí)際執(zhí)行SQL，但仍未必
   是真實(shí)計(jì)劃
   ? 必須要有plan_table
3. SQL TRACE
   ? 需要啟用10046戒者SQL_TRACE
   ? 一般用tkprof看的更清楚些，當(dāng)然10046里本身也有執(zhí)行計(jì)劃信息
4. V$SQL和V$SQL_PLAN
   ? 可以查詢到多個(gè)子游標(biāo)的計(jì)劃信息了，但是看起來(lái)比較費(fèi)勁
5. Enterprise Manager
   ? 可以圖形化顯示執(zhí)行計(jì)劃，但并非所有環(huán)境有EM可用
6. 其他第三方工具
   ? 注意 PL/SQL developer之類工具F5看到的執(zhí)行計(jì)劃未必是真實(shí)的

D:\instantclient_12_2>sqlplus scott/tiger@clonepdb_plug

SQL*Plus: Release 12.2.0.1.0 Production on Fri Jan 12 13:54:31 2018

Copyright (c) 1982, 2016, Oracle. All rights reserved.

Last Successful login time: Fri Jan 12 2018 11:18:59 +08:00

Connected to:
Oracle Database 12c Enterprise Edition Release 12.2.0.1.0 - 64bit Production

SCOTT@clonepdb_plug>set linesize 180
SCOTT@clonepdb_plug>set pagesize 0
SCOTT@clonepdb_plug>alter session set STATISTICS_LEVEL = ALL;

Session altered.

SCOTT@clonepdb_plug>select * from emp where ename='SMITH';
7369 SMITH CLERK 7902 1980-12-17 00:00:00 800 20

SCOTT@clonepdb_plug>select * from table(dbms_xplan.display_cursor(null,null,'allstats'));
SQL_ID cgf95c3k5mszx, child number 0

select * from emp where ename='SMITH'

Plan hash value: 3956160932

| Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers |

| 0 | SELECT STATEMENT | | 1 | | 1 |00:00:00.01 | 8 |
|* 1 | TABLE ACCESS FULL| EMP | 1 | 1 | 1 |00:00:00.01 | 8 |

Predicate Information (identified by operation id):

1 - filter("ENAME"='SMITH')

18 rows selected.

SCOTT@clonepdb_plug>

需要授權(quán)
grant select on v$session to scott;
grant select on v$sql to scott;
grant select on v_$sqlplan to scott;
grant select on v$sql_plan_statistics_all to scott;

1. 基本字段(總是可用的)

Id 執(zhí)行計(jì)劃中每一個(gè)操作(行)的標(biāo)識(shí)符。如果數(shù)字前面帶有星號(hào)，意味著將在隨后提供這行包含的謂詞信息
Operation 對(duì)應(yīng)執(zhí)行的操作。也叫行源操作
Name 操作的對(duì)象名稱
2.查詢優(yōu)化器評(píng)估信息

Rows(E-Rows) 預(yù)估操作返回的記錄條數(shù)
Bytes(E-Bytes) 預(yù)估操作返回的記錄字節(jié)數(shù)
TempSpc 預(yù)估操作使用臨時(shí)表空間的大小
Cost(%CPU) 預(yù)估操作所需的開(kāi)銷。在括號(hào)中列出了CPU開(kāi)銷的百分比。注意這些值是通過(guò)執(zhí)行計(jì)劃計(jì)算出來(lái)的。換句話說(shuō)，父操作的開(kāi)銷包含子操作的開(kāi)銷
Time 預(yù)估執(zhí)行操作所需要的時(shí)間(HH:MM:SS)
3.分區(qū)(僅當(dāng)訪問(wèn)分區(qū)表時(shí)下列字段可見(jiàn))

Pstart 訪問(wèn)的第一個(gè)分區(qū)。如果解析時(shí)不知道是哪個(gè)分區(qū)就設(shè)為KEY,KEY(I),KEY(MC),KEY(OR),KEY(SQ)
Pstop 訪問(wèn)的最后一個(gè)分區(qū)。如果解析時(shí)不知道是哪個(gè)分區(qū)就設(shè)為KEY,KEY(I),KEY(MC),KEY(OR),KEY(SQ)
4.并行和分布式處理(僅當(dāng)使用并行或分布式操作時(shí)下列字段可見(jiàn))

Inst 在分布式操作中，指操作使用的數(shù)據(jù)庫(kù)鏈接的名字
TQ 在并行操作中，用于從屬線程間通信的表隊(duì)列
IN-OUT 并行或分布式操作間的關(guān)系
PQ Distrib 在并行操作中，生產(chǎn)者為發(fā)送數(shù)據(jù)給消費(fèi)者進(jìn)行的分配
5.運(yùn)行時(shí)統(tǒng)計(jì)(當(dāng)設(shè)定參數(shù)statistics_level為all或使用gather_plan_statistics提示時(shí)，下列字段可見(jiàn))

Starts 指定操作執(zhí)行的次數(shù)
A-Rows 操作返回的真實(shí)記錄數(shù)
A-Time 操作執(zhí)行的真實(shí)時(shí)間(HH:MM:SS.FF)
6.I/O 統(tǒng)計(jì)(當(dāng)設(shè)定參數(shù)statistics_level為all或使用gather_plan_statistics提示時(shí)，下列字段可見(jiàn))

Buffers 執(zhí)行期間進(jìn)行的邏輯讀操作數(shù)量
Reads 執(zhí)行期間進(jìn)行的物理讀操作數(shù)量
Writes 執(zhí)行期間進(jìn)行的物理寫(xiě)操作數(shù)量
7.內(nèi)存使用統(tǒng)計(jì)

OMem 最優(yōu)執(zhí)行所需內(nèi)存的預(yù)估值
1Mem 一次通過(guò)(one-pass)執(zhí)行所需內(nèi)存的預(yù)估值
0/1/M 最優(yōu)/一次通過(guò)/多次通過(guò)(multipass)模式操作執(zhí)行的次數(shù)
Used-Mem 最后一次執(zhí)行時(shí)操作使用的內(nèi)存量
Used-Tmp 最后一次執(zhí)行時(shí)操作使用的臨時(shí)空間大小。這個(gè)字段必須擴(kuò)大1024倍才能和其他衡量?jī)?nèi)存的字段一致(比如，32k意味著32MB)
Max-Tmp 操作使用的最大臨時(shí)空間大小。這個(gè)字段必須擴(kuò)大1024倍才能和其他衡量?jī)?nèi)存的字段一致(比如，32k意味著32MB)
例
SCOTT@clonepdb_plug>explain plan for
2 select * from emp e,dept d
3 where e.deptno=d.deptno
4 and e.ename='SMITH';

Explained.

SCOTT@clonepdb_plug>select from table(dbms_xplan.display(null,null,'advanced'));
Plan hash value: 3625962092 -----是指執(zhí)行計(jì)劃轉(zhuǎn)成的hash值,和執(zhí)行效率無(wú)關(guān)，執(zhí)行計(jì)劃的唯一標(biāo)示
/執(zhí)行計(jì)劃部分*/

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

| 0 | SELECT STATEMENT | | 1 | 59 | 4 (0)| 00:00:01 |
| 1 | NESTED LOOPS | | 1 | 59 | 4 (0)| 00:00:01 |
| 2 | NESTED LOOPS | | 1 | 59 | 4 (0)| 00:00:01 |
| 3 | TABLE ACCESS FULL | EMP | 1 | 39 | 3 (0)| 00:00:01 |
| 4 | INDEX UNIQUE SCAN | PK_DEPT | 1 | | 0 (0)| 00:00:01 |
| 5 | TABLE ACCESS BY INDEX ROWID| DEPT | 1 | 20 | 1 (0)| 00:00:01 |

Query Block Name / Object Alias (identified by operation id): --這部分顯示的為查詢塊名和對(duì)象別名

3 - SEL$1 / E@SEL$1 -E@SEL$1，對(duì)應(yīng)到執(zhí)行計(jì)劃中的操作ID為3上，即在表E上的查詢，E為別名，下面類同 SEL$為select 的縮寫(xiě)，位于塊1,相應(yīng)的還有DEL$,INS$,UPD$等
4 - SEL$1 / D@SEL$1
5 - SEL$1 / D@SEL$1

Outline Data --提綱部分，這部分將執(zhí)行計(jì)劃中的圖形化方式以文本形式來(lái)呈現(xiàn)，即轉(zhuǎn)換為提示符方式

/+
BEGIN_OUTLINE_DATA
NLJ_BATCHING(@"SEL$1" "D"@"SEL$1") --指示優(yōu)化器以嵌套循環(huán)關(guān)聯(lián)預(yù)提取的方式訪問(wèn)表
USE_NL(@"SEL$1" "D"@"SEL$1") --使用USE_NL提示，即嵌套循環(huán)
LEADING(@"SEL$1" "E"@"SEL$1" "D"@"SEL$1") --指明前導(dǎo)表
INDEX(@"SEL$1" "D"@"SEL$1" ("DEPT"."DEPTNO"))-指明對(duì)于D上的訪問(wèn)方式為使用索引
FULL(@"SEL$1" "E"@"SEL$1") --指明對(duì)于E上的訪問(wèn)方式為全表掃描 OUTLINE_LEAF(@"SEL$1")
br/>OUTLINE_LEAF(@"SEL$1")
DB_VERSION('12.2.0.1')
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
IGNORE_OPTIM_EMBEDDED_HINTS
END_OUTLINE_DATA
/

Predicate Information (identified by operation id):--謂詞信息部分，在執(zhí)行計(jì)劃中ID帶有星號(hào)的每一行均對(duì)應(yīng)到下面中的一行

3 - filter("E"."ENAME"='SMITH')
4 - access("E"."DEPTNO"="D"."DEPTNO")

Column Projection Information (identified by operation id):--執(zhí)行時(shí)每一步驟所返回的列，下面的不同步驟返回了不同的列

1 - (#keys=0) "E"."EMPNO"[NUMBER,22], "E"."ENAME"[VARCHAR2,10],
"E"."JOB"[VARCHAR2,9], "E"."MGR"[NUMBER,22], "E"."HIREDATE"[DATE,7],
"E"."SAL"[NUMBER,22], "E"."COMM"[NUMBER,22], "E"."DEPTNO"[NUMBER,22],
"D"."DEPTNO"[NUMBER,22], "D"."DNAME"[VARCHAR2,14], "D"."LOC"[VARCHAR2,13]
2 - (#keys=0) "E"."EMPNO"[NUMBER,22], "E"."ENAME"[VARCHAR2,10],
"E"."JOB"[VARCHAR2,9], "E"."MGR"[NUMBER,22], "E"."HIREDATE"[DATE,7],
"E"."SAL"[NUMBER,22], "E"."COMM"[NUMBER,22], "E"."DEPTNO"[NUMBER,22],
"D".ROWID[ROWID,10], "D"."DEPTNO"[NUMBER,22]
3 - "E"."EMPNO"[NUMBER,22], "E"."ENAME"[VARCHAR2,10], "E"."JOB"[VARCHAR2,9],
"E"."MGR"[NUMBER,22], "E"."HIREDATE"[DATE,7], "E"."SAL"[NUMBER,22],
"E"."COMM"[NUMBER,22], "E"."DEPTNO"[NUMBER,22]
4 - "D".ROWID[ROWID,10], "D"."DEPTNO"[NUMBER,22]
5 - "D"."DNAME"[VARCHAR2,14], "D"."LOC"[VARCHAR2,13]

Note--注釋與描述部分，下面的描述中給出了本次SQL語(yǔ)句使用了動(dòng)態(tài)采樣功能

• this is an adaptive plan

64 rows selected.
例
SCOTT@clonepdb_plug>select /+ gather_plan_statistics / * from emp e,dept d
2 where e.deptno=d.deptno
3 and e.ename='SMITH';
--注意此處增加了提示gather_plan_statistics并且該語(yǔ)句被執(zhí)行 等同于alter session set STATISTICS_LEVEL = ALL;
7369 SMITH CLERK 7902 1980-12-17 00:00:00 800 20 20 RESEARCH
DALLAS

SCOTT@clonepdb_plug>select * from table(dbms_xplan.display_cursor(null,null,'iostats last'));--使用display_cursor獲取實(shí)際的執(zhí)行計(jì)劃
SQL_ID 0wb0wwb3621n2, child number 0 --這部分給出了SQL語(yǔ)句的SQL_ID,子游標(biāo)號(hào)以及原始的SQL語(yǔ)句

select /+ gather_plan_statistics / * from emp e,dept d
where e.deptno=d.deptno and e.ename='SMITH'

Plan hash value: 3625962092

| Id | Operation | Name | Starts | E-Rows | A-Rows | A-Time | Buffers |

| 0 | SELECT STATEMENT | | 1 | | 1 |00:00:00.01 | 9 |
| 1 | NESTED LOOPS | | 1 | 1 | 1 |00:00:00.01 | 9 |
| 2 | NESTED LOOPS | | 1 | 1 | 1 |00:00:00.01 | 8 |
| 3 | TABLE ACCESS FULL | EMP | 1 | 1 | 1 |00:00:00.01 | 7 |
| 4 | INDEX UNIQUE SCAN | PK_DEPT | 1 | 1 | 1 |00:00:00.01 | 1 |
| 5 | TABLE ACCESS BY INDEX ROWID| DEPT | 1 | 1 | 1 |00:00:00.01 | 1 |

Predicate Information (identified by operation id):

3 - filter("E"."ENAME"='SMITH')
4 - access("E"."DEPTNO"="D"."DEPTNO")

Note

• this is an adaptive plan
  28 rows selected.

  SCOTT@clonepdb_plug>select * from table(dbms_xplan.display_cursor('6h2msdzb5hj7s',null,'ADVANCED peeked_binds ALLSTATS'));
  SQL_ID 6h2msdzb5hj7s, child number 0

  select /+ gather_plan_statistics / * from emp e,dept d
  where e.deptno=d.deptno and rownum<20 and ename like
  '%I%'

Plan hash value: 2662388210

| Id | Operation | Name | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time | A-Rows | A-Time | Buffers | OMem | 1Mem | O/1/M |

| 0 | SELECT STATEMENT | | 1 | | | 4 (100)| | 4 |00:00:00.02 | 15 | | | |
| 1 | COUNT STOPKEY | | 1 | | | | | 4 |00:00:00.02 | 15 | | | |
| 2 | HASH JOIN | | 1 | 1 | 59 | 4 (0)| 00:00:01 | 4 |00:00:00.02 | 15 | 1055K| 1055K| 1/0/0|
|* 3 | TABLE ACCESS FULL| EMP | 1 | 1 | 39 | 3 (0)| 00:00:01 | 4 |00:00:00.01 | 7 | | | |
| 4 | TABLE ACCESS FULL| DEPT | 1 | 1 | 20 | 1 (0)| 00:00:01 | 3 |00:00:00.01 | 8 | | | |

Query Block Name / Object Alias (identified by operation id):

1 - SEL$1
3 - SEL$1 / E@SEL$1
4 - SEL$1 / D@SEL$1

Outline Data

/+
BEGIN_OUTLINE_DATA
FULL(@"SEL$1" "D"@"SEL$1")
USE_HASH(@"SEL$1" "D"@"SEL$1")IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
DB_VERSION('12.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SEL$1")
br/>IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('12.2.0.1')
DB_VERSION('12.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SEL$1")
LEADING(@"SEL$1" "E"@"SEL$1" "D"@"SEL$1")
END_OUTLINE_DATA
/

Predicate Information (identified by operation id):

1 - filter(ROWNUM<20)
2 - access("E"."DEPTNO"="D"."DEPTNO")
3 - filter(("ENAME" LIKE '%I%' AND "ENAME" IS NOT NULL))

Column Projection Information (identified by operation id):

1 - "E"."DEPTNO"[NUMBER,22], "D"."DEPTNO"[NUMBER,22], "E"."EMPNO"[NUMBER,22], "ENAME"[VARCHAR2,10], "E"."JOB"[VARCHAR2,9],
"E"."MGR"[NUMBER,22], "E"."HIREDATE"[DATE,7], "E"."SAL"[NUMBER,22], "E"."COMM"[NUMBER,22], "D"."LOC"[VARCHAR2,13],
"D"."DEPTNO"[NUMBER,22], "D"."DNAME"[VARCHAR2,14], "D"."LOC"[VARCHAR2,13], "D"."DNAME"[VARCHAR2,14]
2 - (#keys=1) "E"."DEPTNO"[NUMBER,22], "D"."DEPTNO"[NUMBER,22], "E"."EMPNO"[NUMBER,22], "ENAME"[VARCHAR2,10], "E"."JOB"[VARCHAR2,9],
"E"."MGR"[NUMBER,22], "E"."HIREDATE"[DATE,7], "E"."SAL"[NUMBER,22], "E"."COMM"[NUMBER,22], "D"."LOC"[VARCHAR2,13],
"D"."DEPTNO"[NUMBER,22], "D"."DNAME"[VARCHAR2,14], "D"."LOC"[VARCHAR2,13], "D"."DNAME"[VARCHAR2,14]
3 - "E"."EMPNO"[NUMBER,22], "ENAME"[VARCHAR2,10], "E"."JOB"[VARCHAR2,9], "E"."MGR"[NUMBER,22], "E"."HIREDATE"[DATE,7],
"E"."SAL"[NUMBER,22], "E"."COMM"[NUMBER,22], "E"."DEPTNO"[NUMBER,22]
4 - "D"."DEPTNO"[NUMBER,22], "D"."DNAME"[VARCHAR2,14], "D"."LOC"[VARCHAR2,13]

Note

• this is an adaptive plan

66 rows selected.

SCOTT@clonepdb_plug>