手动创建SQL_profile 改变和稳定 SQL 执行计划

原来语句执行计划

SELECT t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id;

---------------------------------------------------------------------------

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

---------------------------------------------------------------------------

|   0 | SELECT STATEMENT   |      |  2500 |   100K|   361   (2)| 00:00:05 |

|*  1 |  HASH JOIN         |      |  2500 |   100K|   361   (2)| 00:00:05 |

|*  2 |   TABLE ACCESS FULL| T1   |  2500 | 75000 |    68   (2)| 00:00:01 |

|   3 |   TABLE ACCESS FULL| T2   | 74790 |   803K|   292   (1)| 00:00:04 |

---------------------------------------------------------------------------

SQL Profiles 中对 Hints有一定的要求。需要经过加工处理

例如手动加入hints  sys.sqlprof_attr (‘USE_NL([email protected]$1 [email protected]$1)‘, ‘INDEX([email protected]$1)‘);

DECLARE

v_hints   SYS.sqlprof_attr;

BEGIN

v_hints :=

sys.sqlprof_attr (‘USE_NL([email protected]$1 [email protected]$1)‘, ‘INDEX([email protected]$1)‘);

DBMS_SQLTUNE.import_sql_profile (

‘select t1.*,t2.owner from t1,t2 where t1.object_name like ‘‘%T1%‘‘ and t1.object_id=t2.object_id‘,

v_hints,

‘SQLPROFILE_NAME1‘,

force_match   => TRUE,

REPLACE       => TRUE);

END;

/* 查看包含的hint */

SELECT attr_val

FROM dba_sql_profiles a, sys.sqlprof$attr b

WHERE a.signature = b.signature AND a.name = ‘SQLPROFILE_NAME1‘;

/* 执行计划已经改变*/

SELECT t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id;

---------------------------------------------------------------------------------------

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

---------------------------------------------------------------------------------------

|   0 | SELECT STATEMENT             |        |  2500 |   100K|  5071   (1)| 00:01:01 |

|   1 |  NESTED LOOPS                |        |       |       |            |          |

|   2 |   NESTED LOOPS               |        |  2500 |   100K|  5071   (1)| 00:01:01 |

|*  3 |    TABLE ACCESS FULL         | T1     |  2500 | 75000 |    68   (2)| 00:00:01 |

|*  4 |    INDEX RANGE SCAN          | T2_IDX |     1 |       |     1   (0)| 00:00:01 |

|   5 |   TABLE ACCESS BY INDEX ROWID| T2     |     1 |    11 |     2   (0)| 00:00:01 |

---------------------------------------------------------------------------------------

那么我们再一次手工修改T1表的统计信息，看看结果如何： 查询计划不变

exec dbms_stats.set_table_stats(‘CQWR‘,‘T1‘,numrows=>5000000);

SELECT t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id;

---------------------------------------------------------------------------------------

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

---------------------------------------------------------------------------------------

|   0 | SELECT STATEMENT             |        |   250K|     9M|   500K  (1)| 01:40:06 |

|   1 |  NESTED LOOPS                |        |       |       |            |          |

|   2 |   NESTED LOOPS               |        |   250K|     9M|   500K  (1)| 01:40:06 |

|*  3 |    TABLE ACCESS FULL         | T1     |   250K|  7324K|   163  (59)| 00:00:02 |

|*  4 |    INDEX RANGE SCAN          | T2_IDX |     1 |       |     1   (0)| 00:00:01 |

|   5 |   TABLE ACCESS BY INDEX ROWID| T2     |     1 |    11 |     2   (0)| 00:00:01 |

---------------------------------------------------------------------------------------

---2对现有的SQL稳定其执行计划。

这里的问题是：稳定一条SQL语句的Hints从哪里来？

简单的sql，没问题，我们可以手工构造，但是复杂的SQL，手工构造相对比较复杂，

同时手工构造的Hints不一定能够保证SQL的执行计划就会稳定。

从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。

从下面可以看到：

SELECT t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id;

select * from table(dbms_xplan.display_cursor(null,null,‘outline‘));

显示的“Outline Data”即是我们稳定SQL执行计划需要的Hints

/*+

BEGIN_OUTLINE_DATA

IGNORE_OPTIM_EMBEDDED_HINTS

OPTIMIZER_FEATURES_ENABLE(‘11.2.0.3‘)

DB_VERSION(‘11.2.0.3‘)

ALL_ROWS

OUTLINE_LEAF(@"SEL$1")

FULL(@"SEL$1" "T1"@"SEL$1")

INDEX(@"SEL$1" "T2"@"SEL$1" ("T2"."OBJECT_ID"))

LEADING(@"SEL$1" "T1"@"SEL$1" "T2"@"SEL$1")

USE_NL(@"SEL$1" "T2"@"SEL$1")

NLJ_BATCHING(@"SEL$1" "T2"@"SEL$1")

END_OUTLINE_DATA

*/

对需要稳定执行计划的SQL，我们所要做的就是如前面所示，将Hints与SQL文本一起创建一个SQL Profile。

一个SQL脚本coe_xfr_sql_profile.sql

下面我们来演示一下整个过程。我们这里要修改执行计划的SQL是：

SELECT t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id;

-----------------------------------------------------------------------------------

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

-----------------------------------------------------------------------------------

|   0 | SELECT STATEMENT   |      |   250K|  9765K|       |   851  (13)| 00:00:11 |

|*  1 |  HASH JOIN         |      |   250K|  9765K|  1144K|   851  (13)| 00:00:11 |

|   2 |   TABLE ACCESS FULL| T2   | 50605 |   543K|       |   155   (1)| 00:00:02 |

|*  3 |   TABLE ACCESS FULL| T1   |   250K|  7080K|       |   150  (65)| 00:00:02 |

-----------------------------------------------------------------------------------

--记住SQL_ID  和 Plan hash value

SQL>select * from table(dbms_xplan.display_cursor(null,null,‘ADVANCED‘));

PLAN_TABLE_OUTPUT

--------------------------------------------------------------------------------

SQL_ID  1atjzcw6x51yp, child number 2

-------------------------------------

SELECT t1.*, t2.owner   FROM t1, t2  WHERE t1.object_name LIKE ‘%T1%‘ AND

t1.object_id = t2.object_id

--我们希望构造的SQL如下

SELECT /*+ modify_sql index(t1) use_nl(t1 t2) */

t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id;

--------------------------------------------------------------------------------------

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

--------------------------------------------------------------------------------------

|   0 | SELECT STATEMENT            |        |   250K|  9765K|   500K  (1)| 01:40:06 |

|   1 |  TABLE ACCESS BY INDEX ROWID| T2     |     1 |    11 |     2   (0)| 00:00:01 |

|   2 |   NESTED LOOPS              |        |   250K|  9765K|   500K  (1)| 01:40:06 |

|*  3 |    TABLE ACCESS FULL        | T1     |   250K|  7080K|   150  (65)| 00:00:02 |

|*  4 |    INDEX RANGE SCAN         | T2_IDX |     1 |       |     1   (0)| 00:00:01 |

--------------------------------------------------------------------------------------

--记住SQL_ID  和 Plan hash value

SQL> select * from table(dbms_xplan.display_cursor(null,null,‘ADVANCED‘));

PLAN_TABLE_OUTPUT

--------------------------------------------------------------------------------

SQL_ID  2ravw02cfdjsg, child number 0

-------------------------------------

SELECT /*+ modify_sql index(t1) use_nl(t1 t2) */       t1.*, t2.owner FROM

t1, t2  WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id

Plan hash value: 3787413387

--分别使用coe_xfr_sql_profile.sql 创建两条语句的SQL_profile

[[email protected] utl]$ cd /home/oracle/sqlt/utl

[[email protected] utl]$ ls

10053.sql                  coe_xfr_sql_profile_1jmk1d81apm35_3787413387.log  flush.sql     sel_aux.sql       sqlthistpurge.sql  sqltmain.sql     xplore

coe_gen_sql_profile.sql    coe_xfr_sql_profile_1jmk1d81apm35_3787413387.sql  plan.sql      sel.sql           sqltimpfo.sql      sqltprofile.sql

coe_load_sql_baseline.sql  coe_xfr_sql_profile.log                           profiler.sql  sqltcdirs.sql     sqltimp.sql        xgram

coe_load_sql_profile.sql   coe_xfr_sql_profile.sql                           pxhcdr.sql    sqlthistfile.sql  sqltlite.sql       xhume

[[email protected] utl]$

[[email protected] utl]$ sqlplus " /as sysdba"

SQL> conn cqwr/cqwr

Connected.

SQL>@coe_xfr_sql_profile.sql

Parameter 1:

SQL_ID (required)

Enter value for 1: 1atjzcw6x51yp

PLAN_HASH_VALUE AVG_ET_SECS

--------------- -----------

2959412835        .138

替换原有语句的SQL文件中内容

将里面h := SYS.SQLPROF_ATTR…

替换成我们想要得到的SQL语句中的那一段。

将这个文件中的force_match从FALSE改成TRUE。

q‘[BEGIN_OUTLINE_DATA]‘,

q‘[IGNORE_OPTIM_EMBEDDED_HINTS]‘,

q‘[OPTIMIZER_FEATURES_ENABLE(‘10.2.0.4‘)]‘,

q‘[ALL_ROWS]‘,

q‘[OUTLINE_LEAF(@"SEL$1")]‘,

q‘[FULL(@"SEL$1" "T1"@"SEL$1")]‘,

q‘[INDEX(@"SEL$1" "T2"@"SEL$1" ("T2"."OBJECT_ID"))]‘,

q‘[LEADING(@"SEL$1" "T1"@"SEL$1" "T2"@"SEL$1")]‘,

q‘[USE_NL(@"SEL$1" "T2"@"SEL$1")]‘,

q‘[END_OUTLINE_DATA]‘);

而后执行原来语句生成的SQL语句

SQL>@coe_xfr_sql_profile_1atjzcw6x51yp_2959412835.sql

---再次执行原始SQL语句我们发现执行计划已经改变

SELECT t1.*, t2.owner

FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id

Plan hash value: 3787413387

--------------------------------------------------------------------------------------

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

--------------------------------------------------------------------------------------

|   0 | SELECT STATEMENT            |        |   250K|  9765K|   500K  (1)| 01:40:06 |

|   1 |  TABLE ACCESS BY INDEX ROWID| T2     |     1 |    11 |     2   (0)| 00:00:01 |

|   2 |   NESTED LOOPS              |        |   250K|  9765K|   500K  (1)| 01:40:06 |

|*  3 |    TABLE ACCESS FULL        | T1     |   250K|  7080K|   150  (65)| 00:00:02 |

|*  4 |    INDEX RANGE SCAN         | T2_IDX |     1 |       |     1   (0)| 00:00:01 |

--------------------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

3 - filter("T1"."OBJECT_NAME" LIKE ‘%T1%‘)

4 - access("T1"."OBJECT_ID"="T2"."OBJECT_ID")

Note

-----

- SQL profile "coe_1atjzcw6x51yp_2959412835" used for this statement

---任务2对现有的SQL稳定其执行计划。比如原有语句已经有hint了

解决办法和上面一样 生成原有语句的HINT 以及需要构造语句的HITN语句

替换其中内容就可以了

select /*+ orig_sql full(t1) full(t2) use_hash(t1 t2) */ t1.*,t2.owner

from t1,t2

where t1.object_name like ‘%T1%‘

and t1.object_id=t2.object_id;

记住 SQL_ID  ft158sw5s22nu

SQL>select * from table(dbms_xplan.display_cursor(null,null,‘ADVANCED‘));

PLAN_TABLE_OUTPUT

--------------------------------------------------------------------------------

SQL_ID  ft158sw5s22nu, child number 0

-------------------------------------

select /*+ orig_sql full(t1) full(t2) use_hash(t1 t2) */ t1.*,t2.owner

from t1,t2          where t1.object_name like ‘%T1%‘          and

t1.object_id=t2.object_id

Plan hash value: 2959412835

SELECT /*+ modify_sql index(t1) use_nl(t1 t2) */

t1.*, t2.owner FROM t1, t2

WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id

记住 SQL_ID  2ravw02cfdjsg

SQL>select * from table(dbms_xplan.display_cursor(null,null,‘ADVANCED‘));

PLAN_TABLE_OUTPUT

--------------------------------------------------------------------------------

SQL_ID  2ravw02cfdjsg, child number 0

-------------------------------------

SELECT /*+ modify_sql index(t1) use_nl(t1 t2) */       t1.*, t2.owner FROM

t1, t2  WHERE t1.object_name LIKE ‘%T1%‘ AND t1.object_id = t2.object_id

Plan hash value: 3787413387

@coe_xfr_sql_profile.sql

原来语句对应生成的SQL

coe_xfr_sql_profile_ft158sw5s22nu_2959412835.sql

想构造的语句生成的SQL

coe_xfr_sql_profile_2ravw02cfdjsg_3787413387.sql

对应的 hint

q‘[BEGIN_OUTLINE_DATA]‘,

q‘[IGNORE_OPTIM_EMBEDDED_HINTS]‘,

q‘[OPTIMIZER_FEATURES_ENABLE(‘10.2.0.4‘)]‘,

q‘[ALL_ROWS]‘,

q‘[OUTLINE_LEAF(@"SEL$1")]‘,

q‘[FULL(@"SEL$1" "T1"@"SEL$1")]‘,

q‘[INDEX(@"SEL$1" "T2"@"SEL$1" ("T2"."OBJECT_ID"))]‘,

q‘[LEADING(@"SEL$1" "T1"@"SEL$1" "T2"@"SEL$1")]‘,

q‘[USE_NL(@"SEL$1" "T2"@"SEL$1")]‘,

q‘[END_OUTLINE_DATA]‘);

替换原有语句的SQL文件中内容

将里面h := SYS.SQLPROF_ATTR…

替换成我们想要得到的SQL语句中的那一段。

将这个文件中的force_match从FALSE改成TRUE。

而后执行原来语句生成的SQL语句

SQL>@coe_xfr_sql_profile_1atjzcw6x51yp_2959412835.sql

select /*+ orig_sql full(t1) full(t2) use_hash(t1 t2) */ t1.*,t2.owner

from t1,t2

where t1.object_name like ‘%T1%‘

and t1.object_id=t2.object_id;

Plan hash value: 3787413387

--------------------------------------------------------------------------------------

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

--------------------------------------------------------------------------------------

|   0 | SELECT STATEMENT            |        |   250K|  9765K|   500K  (1)| 01:40:06 |

|   1 |  TABLE ACCESS BY INDEX ROWID| T2     |     1 |    11 |     2   (0)| 00:00:01 |

|   2 |   NESTED LOOPS              |        |   250K|  9765K|   500K  (1)| 01:40:06 |

|*  3 |    TABLE ACCESS FULL        | T1     |   250K|  7080K|   150  (65)| 00:00:02 |

|*  4 |    INDEX RANGE SCAN         | T2_IDX |     1 |       |     1   (0)| 00:00:01 |

--------------------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

3 - filter("T1"."OBJECT_NAME" LIKE ‘%T1%‘)

4 - access("T1"."OBJECT_ID"="T2"."OBJECT_ID")

Note

-----

- SQL profile "coe_ft158sw5s22nu_2959412835" used for this statement

---------------------------------------------------------------------------------------------------

---------------------------------------------------------------------------------------------------

coe_xfr_sql_profile.sql

SPO coe_xfr_sql_profile.log;
SET DEF ON TERM OFF ECHO ON FEED OFF VER OFF HEA ON LIN 2000 PAGES 100 LONG 8000000 LONGC 800000 TRIMS ON TI OFF TIMI OFF SERVEROUT ON SIZE 1000000 NUMF "" SQLP SQL>;
SET SERVEROUT ON SIZE UNL;
REM
REM $Header: 215187.1 coe_xfr_sql_profile.sql 11.4.4.5 2012/05/02 carlos.sierra $
REM
REM Copyright (c) 2000-2012, Oracle Corporation. All rights reserved.
REM
REM AUTHOR
REM   [email protected]
REM
REM SCRIPT
REM   coe_xfr_sql_profile.sql
REM
REM DESCRIPTION
REM   This script generates another that contains the commands to
REM   create a manual custom SQL Profile out of a known plan from
REM   memory or AWR. The manual custom profile can be implemented
REM   into the same SOURCE system where the plan was retrieved,
REM   or into another similar TARGET system that has same schema
REM   objects referenced by the SQL that generated the known plan.
REM
REM PRE-REQUISITES
REM   1. Oracle Tuning Pack license.
REM
REM PARAMETERS
REM   1. SQL_ID (required)
REM   2. Plan Hash Value for which a manual custom SQL Profile is
REM      needed (required). A list of known plans is presented.
REM      You may choose from list provided or enter a valid phv
REM      from a version of the SQL modified with Hints.
REM
REM EXECUTION
REM   1. Connect into SQL*Plus as user with access to data dictionary.
REM   2. Execute script coe_xfr_sql_profile.sql passing SQL_ID and
REM      plan hash value (parameters can be passed inline or until
REM      requested).
REM
REM EXAMPLE
REM   # sqlplus system
REM   SQL> START coe_xfr_sql_profile.sql [SQL_ID] [PLAN_HASH_VALUE];
REM   SQL> START coe_xfr_sql_profile.sql gnjy0mn4y9pbm 2055843663;
REM   SQL> START coe_xfr_sql_profile.sql gnjy0mn4y9pbm;
REM   SQL> START coe_xfr_sql_profile.sql;
REM
REM NOTES
REM   1. For possible errors see coe_xfr_sql_profile.log
REM   2. If SQLT is installed in SOURCE, you can use instead:
REM      sqlt/utl/sqltprofile.sql
REM   3. Be aware that using DBMS_SQLTUNE requires a license for
REM      Oracle Tuning Pack.
REM
SET TERM ON ECHO OFF;
PRO
PRO Parameter 1:
PRO SQL_ID (required)
PRO
DEF sql_id = ‘&1‘;
PRO
WITH
p AS (
SELECT plan_hash_value
  FROM gv$sql_plan
 WHERE sql_id = TRIM(‘&&sql_id.‘)
   AND other_xml IS NOT NULL
 UNION
SELECT plan_hash_value
  FROM dba_hist_sql_plan
 WHERE sql_id = TRIM(‘&&sql_id.‘)
   AND other_xml IS NOT NULL ),
m AS (
SELECT plan_hash_value,
       SUM(elapsed_time)/SUM(executions) avg_et_secs
  FROM gv$sql
 WHERE sql_id = TRIM(‘&&sql_id.‘)
   AND executions > 0
 GROUP BY
       plan_hash_value ),
a AS (
SELECT plan_hash_value,
       SUM(elapsed_time_total)/SUM(executions_total) avg_et_secs
  FROM dba_hist_sqlstat
 WHERE sql_id = TRIM(‘&&sql_id.‘)
   AND executions_total > 0
 GROUP BY
       plan_hash_value )
SELECT p.plan_hash_value,
       ROUND(NVL(m.avg_et_secs, a.avg_et_secs)/1e6, 3) avg_et_secs
  FROM p, m, a
 WHERE p.plan_hash_value = m.plan_hash_value(+)
   AND p.plan_hash_value = a.plan_hash_value(+)
 ORDER BY
       avg_et_secs NULLS LAST;
PRO
PRO Parameter 2:
PRO PLAN_HASH_VALUE (required)
PRO
DEF plan_hash_value = ‘&2‘;
PRO
PRO Values passed to coe_xfr_sql_profile:
PRO ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
PRO SQL_ID         : "&&sql_id."
PRO PLAN_HASH_VALUE: "&&plan_hash_value."
PRO
SET TERM OFF ECHO ON;
WHENEVER SQLERROR EXIT SQL.SQLCODE;

-- trim parameters
COL sql_id NEW_V sql_id FOR A30;
COL plan_hash_value NEW_V plan_hash_value FOR A30;
SELECT TRIM(‘&&sql_id.‘) sql_id, TRIM(‘&&plan_hash_value.‘) plan_hash_value FROM DUAL;

VAR sql_text CLOB;
VAR sql_text2 CLOB;
VAR other_xml CLOB;
EXEC :sql_text := NULL;
EXEC :sql_text2 := NULL;
EXEC :other_xml := NULL;

-- get sql_text from memory
DECLARE
  l_sql_text VARCHAR2(32767);
BEGIN -- 10g see bug 5017909
  FOR i IN (SELECT DISTINCT piece, sql_text
              FROM gv$sqltext_with_newlines
             WHERE sql_id = TRIM(‘&&sql_id.‘)
             ORDER BY 1, 2)
  LOOP
    IF :sql_text IS NULL THEN
      DBMS_LOB.CREATETEMPORARY(:sql_text, TRUE);
      DBMS_LOB.OPEN(:sql_text, DBMS_LOB.LOB_READWRITE);
    END IF;
    -- removes NUL characters
    l_sql_text := REPLACE(i.sql_text, CHR(00), ‘ ‘);
    -- adds a NUL character at the end of each line
    DBMS_LOB.WRITEAPPEND(:sql_text, LENGTH(l_sql_text) + 1, l_sql_text||CHR(00));
  END LOOP;
  -- if found in memory then sql_text is not null
  IF :sql_text IS NOT NULL THEN
    DBMS_LOB.CLOSE(:sql_text);
  END IF;
EXCEPTION
  WHEN OTHERS THEN
    DBMS_OUTPUT.PUT_LINE(‘getting sql_text from memory: ‘||SQLERRM);
    :sql_text := NULL;
END;
/

SELECT :sql_text FROM DUAL;

-- get sql_text from awr
DECLARE
  l_sql_text VARCHAR2(32767);
  l_clob_size NUMBER;
  l_offset NUMBER;
BEGIN
  IF :sql_text IS NULL OR NVL(DBMS_LOB.GETLENGTH(:sql_text), 0) = 0 THEN
    SELECT sql_text
      INTO :sql_text2
      FROM dba_hist_sqltext
     WHERE sql_id = TRIM(‘&&sql_id.‘)
       AND sql_text IS NOT NULL
       AND ROWNUM = 1;
  END IF;
  -- if found in awr then sql_text2 is not null
  IF :sql_text2 IS NOT NULL THEN
    l_clob_size := NVL(DBMS_LOB.GETLENGTH(:sql_text2), 0);
    l_offset := 1;
    DBMS_LOB.CREATETEMPORARY(:sql_text, TRUE);
    DBMS_LOB.OPEN(:sql_text, DBMS_LOB.LOB_READWRITE);
    -- store in clob as 64 character pieces plus a NUL character at the end of each piece
    WHILE l_offset < l_clob_size
    LOOP
      IF l_clob_size - l_offset > 64 THEN
        l_sql_text := REPLACE(DBMS_LOB.SUBSTR(:sql_text2, 64, l_offset), CHR(00), ‘ ‘);
      ELSE -- last piece
        l_sql_text := REPLACE(DBMS_LOB.SUBSTR(:sql_text2, l_clob_size - l_offset + 1, l_offset), CHR(00), ‘ ‘);
      END IF;
      DBMS_LOB.WRITEAPPEND(:sql_text, LENGTH(l_sql_text) + 1, l_sql_text||CHR(00));
      l_offset := l_offset + 64;
    END LOOP;
    DBMS_LOB.CLOSE(:sql_text);
  END IF;
EXCEPTION
  WHEN OTHERS THEN
    DBMS_OUTPUT.PUT_LINE(‘getting sql_text from awr: ‘||SQLERRM);
    :sql_text := NULL;
END;
/

SELECT :sql_text2 FROM DUAL;
SELECT :sql_text FROM DUAL;

-- validate sql_text
SET TERM ON;
BEGIN
  IF :sql_text IS NULL THEN
    RAISE_APPLICATION_ERROR(-20100, ‘SQL_TEXT for SQL_ID &&sql_id. was not found in memory (gv$sqltext_with_newlines) or AWR (dba_hist_sqltext).‘);
  END IF;
END;
/
SET TERM OFF;

-- get other_xml from memory
BEGIN
  FOR i IN (SELECT other_xml
              FROM gv$sql_plan
             WHERE sql_id = TRIM(‘&&sql_id.‘)
               AND plan_hash_value = TO_NUMBER(TRIM(‘&&plan_hash_value.‘))
               AND other_xml IS NOT NULL
             ORDER BY
                   child_number, id)
  LOOP
    :other_xml := i.other_xml;
    EXIT; -- 1st
  END LOOP;
EXCEPTION
  WHEN OTHERS THEN
    DBMS_OUTPUT.PUT_LINE(‘getting other_xml from memory: ‘||SQLERRM);
    :other_xml := NULL;
END;
/

-- get other_xml from awr
BEGIN
  IF :other_xml IS NULL OR NVL(DBMS_LOB.GETLENGTH(:other_xml), 0) = 0 THEN
    FOR i IN (SELECT other_xml
                FROM dba_hist_sql_plan
               WHERE sql_id = TRIM(‘&&sql_id.‘)
                 AND plan_hash_value = TO_NUMBER(TRIM(‘&&plan_hash_value.‘))
                 AND other_xml IS NOT NULL
               ORDER BY
                     id)
    LOOP
      :other_xml := i.other_xml;
      EXIT; -- 1st
    END LOOP;
  END IF;
EXCEPTION
  WHEN OTHERS THEN
    DBMS_OUTPUT.PUT_LINE(‘getting other_xml from awr: ‘||SQLERRM);
    :other_xml := NULL;
END;
/

-- get other_xml from memory from modified SQL
BEGIN
  IF :other_xml IS NULL OR NVL(DBMS_LOB.GETLENGTH(:other_xml), 0) = 0 THEN
    FOR i IN (SELECT other_xml
                FROM gv$sql_plan
               WHERE plan_hash_value = TO_NUMBER(TRIM(‘&&plan_hash_value.‘))
                 AND other_xml IS NOT NULL
               ORDER BY
                     child_number, id)
    LOOP
      :other_xml := i.other_xml;
      EXIT; -- 1st
    END LOOP;
  END IF;
EXCEPTION
  WHEN OTHERS THEN
    DBMS_OUTPUT.PUT_LINE(‘getting other_xml from memory: ‘||SQLERRM);
    :other_xml := NULL;
END;
/

-- get other_xml from awr from modified SQL
BEGIN
  IF :other_xml IS NULL OR NVL(DBMS_LOB.GETLENGTH(:other_xml), 0) = 0 THEN
    FOR i IN (SELECT other_xml
                FROM dba_hist_sql_plan
               WHERE plan_hash_value = TO_NUMBER(TRIM(‘&&plan_hash_value.‘))
                 AND other_xml IS NOT NULL
               ORDER BY
                     id)
    LOOP
      :other_xml := i.other_xml;
      EXIT; -- 1st
    END LOOP;
  END IF;
EXCEPTION
  WHEN OTHERS THEN
    DBMS_OUTPUT.PUT_LINE(‘getting other_xml from awr: ‘||SQLERRM);
    :other_xml := NULL;
END;
/

SELECT :other_xml FROM DUAL;

-- validate other_xml
SET TERM ON;
BEGIN
  IF :other_xml IS NULL THEN
    RAISE_APPLICATION_ERROR(-20101, ‘PLAN for SQL_ID &&sql_id. and PHV &&plan_hash_value. was not found in memory (gv$sql_plan) or AWR (dba_hist_sql_plan).‘);
  END IF;
END;
/
SET TERM OFF;

-- generates script that creates sql profile in target system:
SET ECHO OFF;
PRO coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..sql.
SET FEED OFF LIN 666 TRIMS ON TI OFF TIMI OFF SERVEROUT ON SIZE 1000000 FOR WOR;
SET SERVEROUT ON SIZE UNL FOR WOR;
SPO OFF;
SPO coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..sql;
DECLARE
  l_pos NUMBER;
  l_clob_size NUMBER;
  l_offset NUMBER;
  l_sql_text VARCHAR2(32767);
  l_len NUMBER;
  l_hint VARCHAR2(32767);
BEGIN
  DBMS_OUTPUT.PUT_LINE(‘SPO coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..log;‘);
  DBMS_OUTPUT.PUT_LINE(‘SET ECHO ON TERM ON LIN 2000 TRIMS ON NUMF 99999999999999999999;‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM $Header: 215187.1 coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..sql 11.4.4.4 ‘||TO_CHAR(SYSDATE, ‘YYYY/MM/DD‘)||‘ carlos.sierra $‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM Copyright (c) 2000-2012, Oracle Corporation. All rights reserved.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM AUTHOR‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   [email protected]‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM SCRIPT‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..sql‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM DESCRIPTION‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   This script is generated by coe_xfr_sql_profile.sql‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   It contains the SQL*Plus commands to create a custom‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   SQL Profile for SQL_ID &&sql_id. based on plan hash‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   value &&plan_hash_value..‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   The custom SQL Profile to be created by this script‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   will affect plans for SQL commands with signature‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   matching the one for SQL Text below.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   Review SQL Text and adjust accordingly.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM PARAMETERS‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   None.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM EXAMPLE‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   SQL> START coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..sql;‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘REM NOTES‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   1. Should be run as SYSTEM or SYSDBA.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   2. User must have CREATE ANY SQL PROFILE privilege.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   3. SOURCE and TARGET systems can be the same or similar.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   4. To drop this custom SQL Profile after it has been created:‘);
  DBMS_OUTPUT.PUT_LINE(‘REM      EXEC DBMS_SQLTUNE.DROP_SQL_PROFILE(‘‘coe_&&sql_id._&&plan_hash_value.‘‘);‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   5. Be aware that using DBMS_SQLTUNE requires a license‘);
  DBMS_OUTPUT.PUT_LINE(‘REM      for the Oracle Tuning Pack.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM   6. If you modified a SQL putting Hints in order to produce a desired‘);
  DBMS_OUTPUT.PUT_LINE(‘REM      Plan, you can remove the artifical Hints from SQL Text pieces below.‘);
  DBMS_OUTPUT.PUT_LINE(‘REM      By doing so you can create a custom SQL Profile for the original‘);
  DBMS_OUTPUT.PUT_LINE(‘REM      SQL but with the Plan captured from the modified SQL (with Hints).‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘WHENEVER SQLERROR EXIT SQL.SQLCODE;‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘VAR signature NUMBER;‘);
  DBMS_OUTPUT.PUT_LINE(‘VAR signaturef NUMBER;‘);
  DBMS_OUTPUT.PUT_LINE(‘REM‘);
  DBMS_OUTPUT.PUT_LINE(‘DECLARE‘);
  DBMS_OUTPUT.PUT_LINE(‘sql_txt CLOB;‘);
  DBMS_OUTPUT.PUT_LINE(‘h       SYS.SQLPROF_ATTR;‘);
  DBMS_OUTPUT.PUT_LINE(‘PROCEDURE wa (p_line IN VARCHAR2) IS‘);
  DBMS_OUTPUT.PUT_LINE(‘BEGIN‘);
  DBMS_OUTPUT.PUT_LINE(‘DBMS_LOB.WRITEAPPEND(sql_txt, LENGTH(p_line), p_line);‘);
  DBMS_OUTPUT.PUT_LINE(‘END wa;‘);
  DBMS_OUTPUT.PUT_LINE(‘BEGIN‘);
  DBMS_OUTPUT.PUT_LINE(‘DBMS_LOB.CREATETEMPORARY(sql_txt, TRUE);‘);
  DBMS_OUTPUT.PUT_LINE(‘DBMS_LOB.OPEN(sql_txt, DBMS_LOB.LOB_READWRITE);‘);
  DBMS_OUTPUT.PUT_LINE(‘-- SQL Text pieces below do not have to be of same length.‘);
  DBMS_OUTPUT.PUT_LINE(‘-- So if you edit SQL Text (i.e. removing temporary Hints),‘);
  DBMS_OUTPUT.PUT_LINE(‘-- there is no need to edit or re-align unmodified pieces.‘);
  l_clob_size := NVL(DBMS_LOB.GETLENGTH(:sql_text), 0);
  l_offset := 1;
  WHILE l_offset < l_clob_size
  LOOP
    l_pos := DBMS_LOB.INSTR(:sql_text, CHR(00), l_offset);
    IF l_pos > 0 THEN
      l_len := l_pos - l_offset;
    ELSE -- last piece
      l_len := l_clob_size - l_pos + 1;
    END IF;
    l_sql_text := DBMS_LOB.SUBSTR(:sql_text, l_len, l_offset);
    /* cannot do such 3 replacement since a line could end with a comment using "--"
    l_sql_text := REPLACE(l_sql_text, CHR(10), ‘ ‘); -- replace LF with SP
    l_sql_text := REPLACE(l_sql_text, CHR(13), ‘ ‘); -- replace CR with SP
    l_sql_text := REPLACE(l_sql_text, CHR(09), ‘ ‘); -- replace TAB with SP
    */
    l_offset := l_offset + l_len + 1;
    IF l_len > 0 THEN
      IF INSTR(l_sql_text, ‘‘‘[‘) + INSTR(l_sql_text, ‘]‘‘‘) = 0 THEN
        l_sql_text := ‘[‘||l_sql_text||‘]‘;
      ELSIF INSTR(l_sql_text, ‘‘‘{‘) + INSTR(l_sql_text, ‘}‘‘‘) = 0 THEN
        l_sql_text := ‘{‘||l_sql_text||‘}‘;
      ELSIF INSTR(l_sql_text, ‘‘‘<‘) + INSTR(l_sql_text, ‘>‘‘‘) = 0 THEN
        l_sql_text := ‘<‘||l_sql_text||‘>‘;
      ELSIF INSTR(l_sql_text, ‘‘‘(‘) + INSTR(l_sql_text, ‘)‘‘‘) = 0 THEN
        l_sql_text := ‘(‘||l_sql_text||‘)‘;
      ELSIF INSTR(l_sql_text, ‘‘‘"‘) + INSTR(l_sql_text, ‘"‘‘‘) = 0 THEN
        l_sql_text := ‘"‘||l_sql_text||‘"‘;
      ELSIF INSTR(l_sql_text, ‘‘‘|‘) + INSTR(l_sql_text, ‘|‘‘‘) = 0 THEN
        l_sql_text := ‘|‘||l_sql_text||‘|‘;
      ELSIF INSTR(l_sql_text, ‘‘‘~‘) + INSTR(l_sql_text, ‘~‘‘‘) = 0 THEN
        l_sql_text := ‘~‘||l_sql_text||‘~‘;
      ELSIF INSTR(l_sql_text, ‘‘‘^‘) + INSTR(l_sql_text, ‘^‘‘‘) = 0 THEN
        l_sql_text := ‘^‘||l_sql_text||‘^‘;
      ELSIF INSTR(l_sql_text, ‘‘‘@‘) + INSTR(l_sql_text, ‘@‘‘‘) = 0 THEN
        l_sql_text := ‘@‘||l_sql_text||‘@‘;
      ELSIF INSTR(l_sql_text, ‘‘‘#‘) + INSTR(l_sql_text, ‘#‘‘‘) = 0 THEN
        l_sql_text := ‘#‘||l_sql_text||‘#‘;
      ELSIF INSTR(l_sql_text, ‘‘‘%‘) + INSTR(l_sql_text, ‘%‘‘‘) = 0 THEN
        l_sql_text := ‘%‘||l_sql_text||‘%‘;
      ELSIF INSTR(l_sql_text, ‘‘‘$‘) + INSTR(l_sql_text, ‘$‘‘‘) = 0 THEN
        l_sql_text := ‘$‘||l_sql_text||‘$‘;
      ELSE
        l_sql_text := CHR(96)||l_sql_text||CHR(96);
      END IF;
      DBMS_OUTPUT.PUT_LINE(‘wa(q‘‘‘||l_sql_text||‘‘‘);‘);
    END IF;
  END LOOP;
  DBMS_OUTPUT.PUT_LINE(‘DBMS_LOB.CLOSE(sql_txt);‘);
  DBMS_OUTPUT.PUT_LINE(‘h := SYS.SQLPROF_ATTR(‘);
  DBMS_OUTPUT.PUT_LINE(‘q‘‘[BEGIN_OUTLINE_DATA]‘‘,‘);
  FOR i IN (SELECT /*+ opt_param(‘parallel_execution_enabled‘, ‘false‘) */
                   SUBSTR(EXTRACTVALUE(VALUE(d), ‘/hint‘), 1, 4000) hint
              FROM TABLE(XMLSEQUENCE(EXTRACT(XMLTYPE(:other_xml), ‘/*/outline_data/hint‘))) d)
  LOOP
    l_hint := i.hint;
    WHILE NVL(LENGTH(l_hint), 0) > 0
    LOOP
      IF LENGTH(l_hint) <= 500 THEN
        DBMS_OUTPUT.PUT_LINE(‘q‘‘[‘||l_hint||‘]‘‘,‘);
        l_hint := NULL;
      ELSE
        l_pos := INSTR(SUBSTR(l_hint, 1, 500), ‘ ‘, -1);
        DBMS_OUTPUT.PUT_LINE(‘q‘‘[‘||SUBSTR(l_hint, 1, l_pos)||‘]‘‘,‘);
        l_hint := ‘   ‘||SUBSTR(l_hint, l_pos);
      END IF;
    END LOOP;
  END LOOP;
  DBMS_OUTPUT.PUT_LINE(‘q‘‘[END_OUTLINE_DATA]‘‘);‘);
  DBMS_OUTPUT.PUT_LINE(‘:signature := DBMS_SQLTUNE.SQLTEXT_TO_SIGNATURE(sql_txt);‘);
  DBMS_OUTPUT.PUT_LINE(‘:signaturef := DBMS_SQLTUNE.SQLTEXT_TO_SIGNATURE(sql_txt, TRUE);‘);
  DBMS_OUTPUT.PUT_LINE(‘DBMS_SQLTUNE.IMPORT_SQL_PROFILE (‘);
  DBMS_OUTPUT.PUT_LINE(‘sql_text    => sql_txt,‘);
  DBMS_OUTPUT.PUT_LINE(‘profile     => h,‘);
  DBMS_OUTPUT.PUT_LINE(‘name        => ‘‘coe_&&sql_id._&&plan_hash_value.‘‘,‘);
  DBMS_OUTPUT.PUT_LINE(‘description => ‘‘coe &&sql_id. &&plan_hash_value. ‘‘||:signature||‘‘ ‘‘||:signaturef||‘‘‘‘,‘);
  DBMS_OUTPUT.PUT_LINE(‘category    => ‘‘DEFAULT‘‘,‘);
  DBMS_OUTPUT.PUT_LINE(‘validate    => TRUE,‘);
  DBMS_OUTPUT.PUT_LINE(‘replace     => TRUE,‘);
  DBMS_OUTPUT.PUT_LINE(‘force_match => FALSE /* TRUE:FORCE (match even when different literals in SQL). FALSE:EXACT (similar to CURSOR_SHARING) */ );‘);
  DBMS_OUTPUT.PUT_LINE(‘DBMS_LOB.FREETEMPORARY(sql_txt);‘);
  DBMS_OUTPUT.PUT_LINE(‘END;‘);
  DBMS_OUTPUT.PUT_LINE(‘/‘);
  DBMS_OUTPUT.PUT_LINE(‘WHENEVER SQLERROR CONTINUE‘);
  DBMS_OUTPUT.PUT_LINE(‘SET ECHO OFF;‘);
  DBMS_OUTPUT.PUT_LINE(‘PRINT signature‘);
  DBMS_OUTPUT.PUT_LINE(‘PRINT signaturef‘);
  DBMS_OUTPUT.PUT_LINE(‘PRO‘);
  DBMS_OUTPUT.PUT_LINE(‘PRO ... manual custom SQL Profile has been created‘);
  DBMS_OUTPUT.PUT_LINE(‘PRO‘);
  DBMS_OUTPUT.PUT_LINE(‘SET TERM ON ECHO OFF LIN 80 TRIMS OFF NUMF "";‘);
  DBMS_OUTPUT.PUT_LINE(‘SPO OFF;‘);
  DBMS_OUTPUT.PUT_LINE(‘PRO‘);
  DBMS_OUTPUT.PUT_LINE(‘PRO COE_XFR_SQL_PROFILE_&&sql_id._&&plan_hash_value. completed‘);
END;
/
SPO OFF;
SET DEF ON TERM ON ECHO OFF FEED 6 VER ON HEA ON LIN 80 PAGES 14 LONG 80 LONGC 80 TRIMS OFF TI OFF TIMI OFF SERVEROUT OFF NUMF "" SQLP SQL>;
SET SERVEROUT OFF;
PRO
PRO Execute coe_xfr_sql_profile_&&sql_id._&&plan_hash_value..sql
PRO on TARGET system in order to create a custom SQL Profile
PRO with plan &&plan_hash_value linked to adjusted sql_text.
PRO
UNDEFINE 1 2 sql_id plan_hash_value
CL COL
PRO
PRO COE_XFR_SQL_PROFILE completed.

时间： 2024-10-07 19:09:54

手动创建SQL_profile 改变和稳定 SQL 执行计划的相关文章

sql_profile 固定SQL执行计划

使用 sql_profile 固定SQL执行计划 Table of Contents 1. 扯蛋 2. 利用SQL PROFILE固定执行计划 2.1. 查看原来语句的执行计划 2.2. 指定SQL使用选择性比较好的索引 2.3. 生成sql profile 2.4. 验证sql profile 是否生效 1 扯蛋 近日 ,每天晚上固定时间,数据库会出现大量的latch:cache buffer chains 等待.经查 看发现原SQL语句走错执行计划,240G的表,进行全表扫描,引起热点块争用

Oracle SQL执行计划基线总结（SQL Plan Baseline）

一.基础概念 Oracle 11g开始,提供了一种新的固定执行计划的方法,即SQL plan baseline,中文名SQL执行计划基线(简称基线),可以认为是OUTLINE(大纲)或者SQL PROFILE的改进版本,基本上它的主要作用可以归纳为如下两个: 1.稳定给定SQL语句的执行计划,防止执行环境或对象统计信息等等因子的改变对SQL语句的执行计划产生影响! 2.减少数据库中出现SQL语句性能退化的概率,理论上不允许一条语句切换到一个比已经执行过的执行计划慢很多的新的执行计划上! 注意:

总有一种SQL执行计划绑定方式合适你

需要绑定SQL执行计划常见的几种情况: SQL执行计划突变,导致数据库性能下降,从历史执行计划找一个合理的,进行绑定. SQL无法使用更优的执行计划,且无历史执行计划,可通过hint手工构造的方式,进行绑定. 某些Bug引起优化器生成较差的执行计划.在bug修复前,进行绑定. ORACLE固定执行计划的3种方式: Oracle 9i使用outline (可跨版本10,11g均可使用) Oracle 10g使用sql profile (11g也可使用) Oracle 11g使用sql plan m

oracle sql 执行计划分析

转自http://itindex.net/detail/45962-oracle-sql-%E8%AE%A1%E5%88%92 一.首先创建表 SQL> show user USER is "RHYS" SQL> create table A(col1 number(4,0),col2 number(4,0), col4 char(30)); create table B(col1 number(4,0),col3 number(4,0), name_b char(30))

替换SQL执行计划

Switching two different SQL Plan with SQL Profile in Oracle... 当SQL是业务系统动态生成的,或者是第三方系统产生的,在数据库层面分析发现性能问题时,可能难以实现及时修改业务程序改善执行计划和性能:但可以在数据库层面找到有问题的SQL,调整和改造该SQL,然后将执行计划应用到原始的SQL语句中,步骤如下: 在数据库层面找到性能问题的SQL相关信息; 重构和优化SQL; 对比旧的和优化后的SQL性能信息; 将最优的SQL执行计划应用到原

DB查询分析器7.01新增的周、月SQL执行计划功能

马根峰              (广东联合电子服务股份有限公司, 广州 510300) 1      引言   中国本土程序员马根峰(CSDN专访马根峰:海量数据处理与分析大师的中国本土程序员)推出的个人作品----万能数据库查询分析器,中文版本DB 查询分析器.英文版本<DB Query Analyzer>.它具有强大的功能.友好的操作界面.良好的操作性.跨越各种数据库平台乃至于EXCEL和文本文件. 你可以通过它 ① 查询ODBC数据源(包括世面上所有的数据库.TXT/CSV文件.EXC

最权威Oracle获取SQL执行计划大全

该文档为根据相关资料整理.总结而成,主要讲解Oracle数据库中,获取SQL语句执行计划的最权威.最正确的方法.步骤,此外,还详细说明了每种方法中可选项的意义及使用方法,以方便大家和自己日常工作中查阅使用,因本人未发现本博客支持附件上传功能,需要PDF文件格式的朋友可向我要,也可到群里下载,转载请注明出处. 1.查询v$sql_plan: SQL> col "Query Plan_Table" format a100 SQL> select id,lpad(' ', 2*(

Oracle查看SQL执行计划的方式

Oracle查看SQL执行计划的方式   获取Oracle sql执行计划并查看执行计划,是掌握和判断数据库性能的基本技巧.下面案例介绍了多种查看sql执行计划的方式: 基本有以下几种方式: 1.通过sql_trace初始化参数 2.通过Autotrace 3.通过explain plan 4.通过dbms_xplan.display_cursor 5.通过dbms_xplan.display_awr 6.通过10046事件 1.通过explain plan 工具 12:24:00 [email

查看Oracle SQL执行计划的常用方式

在查看SQL执行计划的时候有很多方式 我常用的方式有三种 SQL> explain plan for 2 select * from scott.emp where ename='KING'; 已解释. 第一种 最常用的 SQL> select * from table(dbms_xplan.display); Plan hash value: 3956160932 ----------------------------------------------------------------