oracle---bulk collect 的用法

通过bulk collect减少loop处理的开销

采用bulk collect可以将查询结果一次性地加载到collections中。
而不是通过cursor一条一条地处理。
可以在select into,fetch into,returning into语句使用bulk collect。
注意在使用bulk collect时，所有的into变量都必须是collections.
-------------------------------------------------------------------------------------------

简单举个例：

 

CREATE OR REPLACE TYPE NODETYPE AS OBJECT
(
  SEQ       NUMBER(10), 
  NODE_CODE VARCHAR2(30)
)
/

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

CREATE OR REPLACE TYPE NODETYPE_LIST AS TABLE OF NODETYPE
/

 

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

FUNCTION GET_QUARTER_OF_MONTH_LIST(P_I_DATE IN VARCHAR2)
    RETURN NODETYPE_LIST IS
    V_YEAR  VARCHAR2(4);
    V_MONTH VARCHAR2(2);
    RESULT  NODETYPE_LIST;
  BEGIN
    V_YEAR  := SUBSTR(P_I_DATE, 1, 4);
    V_MONTH := SUBSTR(P_I_DATE, 5, 2);
    RESULT  := NODETYPE_LIST();
 
    CASE
      WHEN V_MONTH IN ('01', '02', '03') THEN
     
        SELECT NODETYPE(ROWNUM,
                        TO_CHAR(LAST_DAY(TO_DATE(F_DATE, 'YYYYMMDD')),
                                'YYYYMMDD')) BULK COLLECT
          INTO RESULT
          FROM (SELECT V_YEAR || '01' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '02' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '03' || '01' AS F_DATE FROM DUAL);
     
      WHEN V_MONTH IN ('04', '05', '06') THEN
        SELECT NODETYPE(ROWNUM,
                        TO_CHAR(LAST_DAY(TO_DATE(F_DATE, 'YYYYMMDD')),
                                'YYYYMMDD')) BULK COLLECT
          INTO RESULT
          FROM (SELECT V_YEAR || '04' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '05' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '06' || '01' AS F_DATE FROM DUAL);
      WHEN V_MONTH IN ('07', '08', '09') THEN
        SELECT NODETYPE(ROWNUM,
                        TO_CHAR(LAST_DAY(TO_DATE(F_DATE, 'YYYYMMDD')),
                                'YYYYMMDD')) BULK COLLECT
          INTO RESULT
          FROM (SELECT V_YEAR || '07' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '08' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '09' || '01' AS F_DATE FROM DUAL);
      WHEN V_MONTH IN ('10', '11', '12') THEN
        SELECT NODETYPE(ROWNUM,
                        TO_CHAR(LAST_DAY(TO_DATE(F_DATE, 'YYYYMMDD')),
                                'YYYYMMDD')) BULK COLLECT
          INTO RESULT
          FROM (SELECT V_YEAR || '10' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '11' || '01' AS F_DATE
                  FROM DUAL
                UNION ALL
                SELECT V_YEAR || '12' || '01' AS F_DATE FROM DUAL);
    END CASE;
 
    RETURN RESULT;
  END;

/

-----------查询结果

SELECT NODE_CODE
                  FROM TABLE(GET_QUARTER_OF_MONTH_LIST(‘20121015’))；

 

---------结果为：

20121031

20121130

20121231

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

通过上面的测试和分析，我们可以看到Bulk
Collect批查询在某种程度上可以提高查询效率，它首先将所需数据读入内存，然后再统计分析，这样就可以提高查询效率。但是，如果Oracle数据库的内存较小，Shared
Pool Size不足以保存Bulk Collect批查询结果，那么该方法需要将Bulk Collect的集合结果保存在磁盘上，在这种情况下，Bulk
Collect方法的效率反而不如其他两种方法，有兴趣的读者可以进一步测试。

另外，除了Bulk Collect批查询外，我们还可以使用FORALL语句来实现批插入、删除和更新，这在大批量数据操作时可以显著提高执行效率。

 

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/26423870/viewspace-754634/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/26423870/viewspace-754634/