oracle中查询表重复字段记录时的效率问题...

查询表中重复字段记录的效率问题...

--目标：取出在t_tr中m_no,sq,p_roll,npl重复的字段，t_tr表记录数5000条

--方法一
--用时： 00: 00: 29.09
select b.m_no,b.sq,b.p_roll,b.npl,b.tr_date from t_tr b,
(select count(*),m_no,sq,p_roll,npl
from t_tr
group by m_no,sq,p_roll,npl
having count(*)>1) a
where b.m_no=a.m_no and b.sq=a.sq and b.p_roll=a.p_roll and b.npl=a.npl
order by b.tr_date

--方法二
--用时：n秒，最后只好卡叉掉
select m_no,sq,p_roll,npl,tr_date
from t_tr a
where a.rowid!=(select max(rowid)
   from t_tr b
   where a.m_no=b.m_no and a.sq=b.sq and a.p_roll=b.p_roll and a.npl=b.npl)

--方法三
--用时：5秒，
--使用oracle的分析函数

select M_NO,SQ,PT,P_ROLL,P_YDS,P_KGS,tr_date,dept from TR
where (M_NO,SQ,nvl(PT,'<NULL>'),p_roll) in
(select M_NO,SQ,Nvl(PT,'<NULL>'),P_ROLL from
 (select M_NO,SQ,PT,P_ROLL,Row_Number() over (partition By M_NO,SQ,PT,P_ROLL order by M_NO,SQ,PT,P_ROLL) rn
    from tr order by M_NO,SQ,PT,P_ROLL) t
  where rn>1)
  and tr_date>to_date('2004-04-01','yyyy-mm-dd')

小结：合理的使用oracle的分析函数可以很大的提高sql的执行效率。