20.       用表连接替换EXISTS

     通常来说 , 采用表连接的方式比EXISTS更有效率

      SELECT ENAME

      FROM EMP E

      WHERE EXISTS (SELECT ‘X’

                      FROM DEPT

                      WHERE DEPT_NO = E.DEPT_NO

                      AND DEPT_CAT = ‘A’);

     (更高效)

      SELECT ENAME

      FROM DEPT D,EMP E

      WHERE E.DEPT_NO = D.DEPT_NO

      AND DEPT_CAT = ‘A’ ;

(译者按: 在RBO的情况下,前者的执行路径包括FILTER,后者使用NESTED LOOP)

21.       用EXISTS替换DISTINCT

当提交一个包含一对多表信息(比如部门表和雇员表)的查询时,避免在SELECT子句中使用DISTINCT. 一般可以考虑用EXIST替换

例如:

低效:

    SELECT DISTINCT DEPT_NO,DEPT_NAME

    FROM DEPT D,EMP E

    WHERE D.DEPT_NO = E.DEPT_NO

高效:

    SELECT DEPT_NO,DEPT_NAME

    FROM DEPT D

    WHERE EXISTS ( SELECT ‘X’

                    FROM EMP E

                    WHERE E.DEPT_NO = D.DEPT_NO);

  EXISTS 使查询更为迅速,因为RDBMS核心模块将在子查询的条件一旦满足后,立刻返回结果.

22.       识别’低效执行’的SQL语句

用下列SQL工具找出低效SQL:

SELECT EXECUTIONS , DISK_READS, BUFFER_GETS,

        ROUND((BUFFER_GETS-DISK_READS)/BUFFER_GETS,2) Hit_radio,

        ROUND(DISK_READS/EXECUTIONS,2) Reads_per_run,

        SQL_TEXT

FROM   V$SQLAREA

WHERE  EXECUTIONS>0

AND     BUFFER_GETS > 0

AND (BUFFER_GETS-DISK_READS)/BUFFER_GETS < 0.8

ORDER BY 4 DESC;

(译者按:

[1] [2]  下一页