日期:2014-05-16  浏览次数:21149 次

这个语句还能不能优化呀?谢谢了。
SQL code

select distinct *
  from (select /*+ leading(T_CARDINFO) INDEX(T_CARDINFO CI_XM_INDEX) INDEX(T_CARDTIME T_CHECKTIME_INDEX) */
         T_CARDINFO.*, rownum rn
          from T_CARDINFO, T_CARDTIME
         where CI_SFZH = CT_SFZH
           and exists (select 'x'
                  from T_CARDTIME
                 where CI_SFZH = CT_SFZH
                   and CT_YZSJ >= to_date('20010501', 'YYYYMMDD')
                   and CT_YZSJ <= to_date('20110921', 'YYYYMMDD'))
           and CI_XM = '李明'
           and CT_CHECK_USER like '800300020600%'
           and rownum < 21) t
 WHERE t.rn >= 1


数据量在500万左右,查询速度太慢。能优化吗??谢谢了

------解决方案--------------------
CT_CHECK_USER 这个字段有索引么?
看执行计划呗。
------解决方案--------------------
先分析每张表,再看执行计划,数据量500万是指哪张表?还是结果是500万.
分析表sql:analyze table T_CARDINFO compute statistics;
DBMS_STATS.gather_table_stats('system', T_CARDINFO');
执行计划sql:
explain plan for select v from t11 where v in (select v from t22);
select * from table(dbms_xplan.display());

------解决方案--------------------
你这个SQL写的有些问题啊,为啥T_CARDTIME关联了两次,查询的结果两次?
写成下面的不行吗?
SQL code
SELECT /*+ leading(T_CARDINFO) INDEX(T_CARDINFO CI_XM_INDEX) INDEX(T_CARDTIME T_CHECKTIME_INDEX) */
DISTINCT T_CARDINFO.*
  FROM T_CARDINFO
 WHERE EXISTS (SELECT 'x'
          FROM T_CARDTIME
         WHERE CI_SFZH = CT_SFZH
           AND CT_YZSJ >= TO_DATE('20010501', 'YYYYMMDD')
           AND CT_YZSJ <= TO_DATE('20110921', 'YYYYMMDD')
           AND CT_CHECK_USER LIKE '800300020600%')
   AND CI_XM = '李明'
   AND ROWNUM < 21 ;

------解决方案--------------------
语句套的层太多。完全减少层数。可以不用distinct,把表的别名带到字段中去。