网上都说union all效率比or快，可是实际使用过程中发现or比union all快，sql语句如下，使用or查询耗时1s多，union all耗时2s多，表数据大概有8千万左右，是按天分区的。
union all 的sql语句1：
select * from t_m_train_record t where t.stunum='5111786145333504' and t.traintime between to_date('20180710122954','yyyymmddhh24miss') and to_date('20180710153940','yyyymmddhh24miss') union all select * from t_m_train_record t where t.stunum='5111786145333504' and t.traintime between to_date('20180711141455','yyyymmddhh24miss') and to_date('20180711161920','yyyymmddhh24miss');
or 的sql语句2：
select * from t_m_train_record t where t.stunum='5111786145333504' and t.traintime between to_date('20180710122954','yyyymmddhh24miss') and to_date('20180710153940','yyyymmddhh24miss') union all select * from t_m_train_record t where t.stunum='5111786145333504' and t.traintime between to_date('20180711141455','yyyymmddhh24miss') and to_date('20180711161920','yyyymmddhh24miss');
不考虑走索引优化之类的。从理论上or是要快一点，因为or只扫描一遍表， union要两次。 但是如果考虑索引的话union all可以比or要快
这个要看具体执行计划和表数据的分布情况
or的那句sql贴错了，两句是一样的
select * from t where a=xx or a=yy
select * from t where a=xx
union all
select * from t where a=yy
应该union all 会快一点
select * from t where (a=xx or a=yy) and b=zz
select * from t where a=xx and b=zz
union all
select * from t where a=yy and b=zz
这种可能or会快一点
select * from t_m_train_record t where t.stunum='5111786145333504' and( t.traintime between to_date('20180710122954','yyyymmddhh24miss') and to_date('20180710153940','yyyymmddhh24miss') or t.traintime between to_date('20180711141455','yyyymmddhh24miss') and to_date('20180711161920','yyyymmddhh24miss')); or的sql语句代码贴错了
通常，
如果查询上有索引，union all比or快，因为前者会利用索引查找，后者不会；
如果查询上没有索引，or 比union快，因为前者查询引擎会一次性完成指令分析
但是，
具体哪个更快，与数据的体量，分区的方式，查询数据落入的区间都相关