Oracle_dbms_shared_pool.purge 包的使用
自10.2.0.4 开始oracle引入一种方法，它可以删除指定的缓存在库缓存中的shared cursor,它可以让oracle 在执行目标SQL时使用硬解析。
现在，我们来测试一下使用DBMS_SHARED_POOL.PURGE.执行一下SQL:
SQL> SELECT SQL_TEXT,SQL_ID,VERSION_COUNT ,executions,ADDRESS,HASH_VALUE FROM V$SQLAREA WHERE SQL_TEXT LIKE 'SELECT COUNT(*) FROM SDXJ.%';
SQL_TEXT SQL_ID VERSION_COUNT EXECUTIONS ADDRESS HASH_VALUE
---------------------------------------------------------------------- -------------------- ------------- ---------- -------------------- --------------
SELECT COUNT(*) FROM SDXJ.T1 WHERE OBJECT_ID BETWEEN 999 AND 60000 6d2d6y357kfws 1 1 B8BD275C 3396942744
SELECT COUNT(*) FROM SDXJ.t2 WHERE OBJECT_ID BETWEEN 999 AND 600000 3j7bxvb0qusb9 2 3 B4C6E11C 3245171049
SELECT COUNT(*) FROM SDXJ.t2 WHERE OBJECT_ID BETWEEN 989 AND 600000 ccw4344g2uvmg 2 3 B4BDD76C 506293871
SELECT COUNT(*) FROM SDXJ.T2 WHERE OBJECT_ID BETWEEN :x AND :y a6fkgvf7s34vd 2 5 B8FD0970 2407633773
---查看执行计划
Plan hash value: 1410530761
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | | | 46 (100)| |
| 1 | SORT AGGREGATE | | 1 | 5 | | |
|* 2 | FILTER | | | | | |
|* 3 | INDEX FAST FULL SCAN| IDX_T1 | 54911 | 268K| 46 (3)| 00:00:01
---------------------------------------------------------------------------------
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
1 - SEL$1
3 - SEL$1 / T2@SEL$1
Outline Data
------------
/*+
BEGIN_OUTLINE_DATA
IGNORE_OPTIM_EMBEDDED_HINTS
OPTIMIZER_FEATURES_ENABLE('11.2.0.1')
DB_VERSION('11.2.0.1')
ALL_ROWS
OUTLINE_LEAF(@"SEL$1")
INDEX_FFS(@"SEL$1" "T2"@"SEL$1" ("T2"."OBJECT_ID"))
END_OUTLINE_DATA
*/
Peeked Binds (identified by position):
--------------------------------------
1 - :X (NUMBER): 999
2 - :Y (NUMBER): 60000
Predicate Information (identified by operation id):
---------------------------------------------------
2 - filter(:X<=:Y)
3 - filter(("OBJECT_ID"<=:Y AND "OBJECT_ID">=:X))
Column Projection Information (identified by operation id):
-----------------------------------------------------------
1 - (#keys=0) COUNT(*)[22]
Note
-----
- SQL plan baseline SQL_PLAN_8c75u4j38bvqy8856a46c used for this statement
已选择56行。
将上述查询到的列address 和 hash-value 拼成一个字符串（中间用逗号分隔）后作为第一个参数传入包，第二个参数为常量:c(表示清除的是shared cursor).
--执行的sql
SQL> exec DBMS_SHARED_POOL.PURGE('B8FD0970,2407633773','c');
PL/SQL 过程已成功完成。
--再次查询结果
SQL> SELECT SQL_TEXT,SQL_ID,VERSION_COUNT ,executions,ADDRESS,HASH_VALUE FROM V$SQLAREA WHERE SQL_TEXT LIKE 'SELECT COUNT(*) FROM SDXJ.%';
SQL_TEXT SQL_ID VERSION_COUNT EXECUTIONS ADDRESS HASH_VALUE
---------------------------------------------------------------------- -------------------- ------------- ---------- -------------------- --------------
SELECT COUNT(*) FROM SDXJ.T1 WHERE OBJECT_ID BETWEEN 999 AND 60000 6d2d6y357kfws 1 1 B8BD275C 3396942744
SELECT COUNT(*) FROM SDXJ.t2 WHERE OBJECT_ID BETWEEN 999 AND 600000 3j7bxvb0qusb9 2 3 B4C6E11C 3245171049
SELECT COUNT(*) FROM SDXJ.t2 WHERE OBJECT_ID BETWEEN 989 AND 600000 ccw4344g2uvmg 2 3 B4BDD76C 506293871
SELECT COUNT(*) FROM SDXJ.T2 WHERE OBJECT_ID BETWEEN :x AND :y a6fkgvf7s34vd 2 5 B8FD0970 2407633773
--这行消失了。
----- 未完待续