查看真实的执行计划 绑定变量对执行计划的影响--“绑定变量窥探”

--#####################################################
--####     AWR执行计划                             ####
--#####################################################
SELECT TF.* FROM DBA_HIST_SQLTEXT HT, TABLE(DBMS_XPLAN.DISPLAY_AWR(HT.SQL_ID, NULL, NULL, ‘all‘)) TF WHERE HT.SQL_TEXT LIKE ‘%mac%‘;  

--#####################################################
--####     dbms_xplan.display_cursor方式查看       ####
--#####################################################
SQL> conn cqwr/cqwr

SELECT T_LTE_CARD_INFO.PK, 
       T_LTE_CARD_INFO.IMSI, 
       T_LTE_CARD_INFO.ICCID, 
       T_LTE_CARD_INFO.PIN1, 
       T_LTE_CARD_INFO.PIN2, 
       T_LTE_CARD_INFO.PUK1, 
       T_LTE_CARD_INFO.PUK2, 
       T_LTE_CARD_INFO.KI, 
       T_LTE_CARD_INFO.EKI, 
       T_LTE_CARD_INFO.STATUS 
  FROM T_LTE_CARD_INFO SAMPLE (5), 
       T_HLR_MSISDN 
 WHERE T_HLR_MSISDN.HLR_PK = ‘HLR12‘ 
   AND T_LTE_CARD_INFO.SUB_IMSI = ‘46002‘ 
   AND T_LTE_CARD_INFO.MSISDN = T_HLR_MSISDN.PK 
   AND ROWNUM <= 5

select A.SQL_ID,A.HASH_VALUE,A.CHILD_NUMBER,A.SQL_TEXT from v$sql a where A.SQL_TEXT like ‘SELECT T_LTE_CARD_INFO.PK,%‘ 
-- 0jq118yf3w76u    2621316314    0

--传入HASH_VALUE
select * from table(dbms_xplan.display_cursor(2621316314,0,‘ADVANCED‘));

--传入SQL_ID
---为了获取缓存库中的执行计划，可以直接查询动态性能视图v$sql_plan和v$sql_plan_statistics_all等，但更方便的方法是：
select t.* from v$sql s,table(dbms_xplan.display_cursor(s.sql_id,s.child_number,‘allstats last‘)) t where s.sql_id = ‘0jq118yf3w76u‘;

---当前查看的方式如下
SQL> select * from table(dbms_xplan.display_cursor);
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------
SQL_ID  b4vp4vx8q5jny, child number 0
-------------------------------------
SELECT T_LTE_CARD_INFO.PK,        T_LTE_CARD_INFO.IMSI,        T_LTE_CARD_INFO.ICCID,
   T_LTE_CARD_INFO.PIN1,        T_LTE_CARD_INFO.PIN2,        T_LTE_CARD_INFO.PUK1,
T_LTE_CARD_INFO.PUK2,        T_LTE_CARD_INFO.KI,        T_LTE_CARD_INFO.EKI,
T_LTE_CARD_INFO.STATUS   FROM T_LTE_CARD_INFO SAMPLE (5),        T_HLR_MSISDN  WHERE
T_HLR_MSISDN.HLR_PK = ‘HLR12‘    AND T_LTE_CARD_INFO.SUB_IMSI = ‘46002‘    AND
T_LTE_CARD_INFO.MSISDN = T_HLR_MSISDN.PK    AND ROWNUM <= 5

Plan hash value: 1908090178


PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------------------
| Id  | Operation                      | Name            | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT               |                 |       |       |     8 (100)|          |
|*  1 |  COUNT STOPKEY                 |                 |       |       |            |          |
|   2 |   TABLE ACCESS BY INDEX ROWID  | T_LTE_CARD_INFO |     6 |  1308 |     6   (0)| 00:00:01 |
|   3 |    NESTED LOOPS                |                 |     6 |  1542 |     8   (0)| 00:00:01 |
|   4 |     TABLE ACCESS BY INDEX ROWID| T_HLR_MSISDN    |    38 |  1482 |     2   (0)| 00:00:01 |
|*  5 |      INDEX RANGE SCAN          | IND_HLR_PK      |       |       |     1   (0)| 00:00:01 |
|*  6 |     INDEX RANGE SCAN           | IND_MS_SU       |     6 |       |     5   (0)| 00:00:01 |
--------------------------------------------------------------------------------------------------

PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter(ROWNUM<=5)
   5 - access("T_HLR_MSISDN"."HLR_PK"=‘HLR12‘)
   6 - access("T_LTE_CARD_INFO"."MSISDN"="T_HLR_MSISDN"."PK" AND
              "T_LTE_CARD_INFO"."SUB_IMSI"=‘46002‘)
       filter(ORA_HASH(ROWID,0,2038870484,‘SYS_SAMPLE‘,0)<214748365)
32 rows selected.

----查看执行计划中的 获取a-rows信息 E ROWS 预估的行数-----
----需要alter session set statistics_level=all 或 /*+ gather_plan_statistics */

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

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

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


PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------
SQL_ID  b4vp4vx8q5jny, child number 1
-------------------------------------
SELECT T_LTE_CARD_INFO.PK,        T_LTE_CARD_INFO.IMSI,        T_LTE_CARD_INFO.ICCID,
T_LTE_CARD_INFO.PIN1,        T_LTE_CARD_INFO.PIN2,        T_LTE_CARD_INFO.PUK1,
T_LTE_CARD_INFO.PUK2,        T_LTE_CARD_INFO.KI,        T_LTE_CARD_INFO.EKI,
T_LTE_CARD_INFO.STATUS   FROM T_LTE_CARD_INFO SAMPLE (5),        T_HLR_MSISDN  WHERE
T_HLR_MSISDN.HLR_PK = ‘HLR12‘    AND T_LTE_CARD_INFO.SUB_IMSI = ‘46002‘    AND
T_LTE_CARD_INFO.MSISDN = T_HLR_MSISDN.PK    AND ROWNUM <= 5

Plan hash value: 1908090178


PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------
| Id  | Operation                      | Name            | Starts | E-Rows | A-Rows |   A-Time   | Buffers |
------------------------------------------------------------------------------------------------------------
|*  1 |  COUNT STOPKEY                 |                 |      1 |        |      5 |00:00:00.01 |      19 |
|   2 |   TABLE ACCESS BY INDEX ROWID  | T_LTE_CARD_INFO |      1 |      6 |      5 |00:00:00.01 |      19 |
|   3 |    NESTED LOOPS                |                 |      1 |      6 |      8 |00:00:00.01 |      15 |
|   4 |     TABLE ACCESS BY INDEX ROWID| T_HLR_MSISDN    |      1 |     38 |      3 |00:00:00.01 |       3 |
|*  5 |      INDEX RANGE SCAN          | IND_HLR_PK      |      1 |        |      3 |00:00:00.01 |       2 |
|*  6 |     INDEX RANGE SCAN           | IND_MS_SU       |      3 |      6 |      5 |00:00:00.01 |      12 |
------------------------------------------------------------------------------------------------------------
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter(ROWNUM<=5)
   5 - access("T_HLR_MSISDN"."HLR_PK"=‘HLR12‘)
   6 - access("T_LTE_CARD_INFO"."MSISDN"="T_HLR_MSISDN"."PK" AND
              "T_LTE_CARD_INFO"."SUB_IMSI"=‘46002‘)
       filter(ORA_HASH(ROWID,0,1153052846,‘SYS_SAMPLE‘,0)<214748365)

31 rows selected.

--#####################################################
--####     绑定变量对执行计划的影响  “绑定变量窥探”      ####
--#####################################################
SQL> CREATE TABLE t1 AS  SELECT * FROM DBA_OBJECTS ;
   
SQL> create index idx_t1 on t1 (object_id);
---收集统计信息
BEGIN
   sys.DBMS_STATS.gather_table_stats (
      ownname            => ‘CQWR‘,
      tabname            => ‘T1‘,
      estimate_percent   => NULL,
      method_opt         => ‘for all indexed columns size auto‘,
      cascade            => TRUE);
END;

--先来看不执行绑定变量时候的执行计划
SQL> set wrap off
SQL> select count(*) from  t1 where object_id between 999 and 1000;

  COUNT(*)
----------
         2


SQL> select count(*) from  t1 where object_id between 0 and 50000;

  COUNT(*)
----------
     47490

SQL> select * from table(dbms_xplan.display_cursor);

PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
SQL_ID  3qwssaf8f9kgt, child number 0
-------------------------------------
select count(*) from  t1 where object_id between 0 and 50000

Plan hash value: 2101382132

--------------------------------------------------------------------------------
| Id  | Operation             | Name   | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT      |        |       |       |    27 (100)|          |
|   1 |  SORT AGGREGATE       |        |     1 |     5 |            |          |

PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
|*  2 |   INDEX FAST FULL SCAN| IDX_T1 | 46783 |   228K|    27   (4)| 00:00:01 |
--------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - filter(("OBJECT_ID"<=50000 AND "OBJECT_ID">=0))
19 rows selected.

和我们预期的一致  INDEX RANGE SCAN  和  INDEX FAST FULL SCAN

我们使用绑定变量的方式来执行SQL

SQL> var x number
SQL> var y number
SQL> exec :x :=99
PL/SQL procedure successfully completed.

SQL> exec :y :=100
PL/SQL procedure successfully completed.

SQL> select count(*) from  t1 where object_id between :x and :y ;

  COUNT(*)
----------
         2

SQL> alter session set statistics_level=all ;

SQL> exec :x :=0
PL/SQL procedure successfully completed.
SQL> exec :y :=50000
PL/SQL procedure successfully completed.

SQL> select count(*) from  t1 where object_id between :x and :y ;

  COUNT(*)
----------
     47490

----可以发现 两次执行计划都是走的 INDEX RANGE SCAN  

因为  Peeked Binds的值还是 99 到100
也就是oracle在处理有绑定变量的时候只有在硬解析的时候才去窥探一下该SQL的绑定变量。
我们可以通过一些无关紧要的DDL语句来实现硬解析。

SQL>ALTER SYSTEM FLUSH SHARED_POOL ;

SQL> select count(*) from  t1 where object_id between :x and :y ;

  COUNT(*)
----------
     47490


在更换一个值 结果又会走INDEX FAST FULL SCAN 以下就略
SQL> exec :x :=0
PL/SQL procedure successfully completed.
SQL> exec :y :=5

可以设置这个参数来跳过 但是这个事内部参数 还是不稳定会有 BUG:4567767 Peeked Binds 还是会发生
alter system set "_optim_peek_user_binds"=true scope=spfile ;