mysqltimestamp和long存储时间效率比较_MySQL

show create table 20130107date;

CREATE TABLE `20130107date` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `c_date` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
  `c_date_long` int(20) NOT NULL,
  `idx_date` timestamp NOT NULL DEFAULT &＃39;0000-00-00 00:00:00&＃39;,
  `idx_date_long` int(11) NOT NULL,
  PRIMARY KEY (`id`),
  KEY `20130107date_idx_date` (`idx_date`),
  KEY `20130107date_idx_long` (`idx_date_long`)
) ENGINE=InnoDB

里面有90w数据,都是随机的时间.
先看没有索引的全表扫描

1 :

select COUNT(*) from 20130107date
where c_date BETWEEN DATE(&＃39;20110101&＃39;) and DATE(&＃39;20110102&＃39;)

这个需要1.54s

2:

select COUNT(*) from 20130107date
where c_date_long BETWEEN UNIX_TIMESTAMP(&＃39;20110101&＃39;) and UNIX_TIMESTAMP(&＃39;20110102&＃39;)

这个是2.3s

但是可以这样搞
3 :

select UNIX_TIMESTAMP(&＃39;20110101&＃39;),UNIX_TIMESTAMP(&＃39;20110102&＃39;);

得到结果1293811200和1293897600

然后

select COUNT(*) from 20130107date
where c_date_long BETWEEN 1293811200 and 1293897600；

发现变成了0.61s
1和2的差距还可以说是比较int和比较timestamp的差距,那么2和3的差距呢?难道多出来的时间是每一条记录都要evaluate UNIX_TIMESTAMP(‘20110102’)?

然后用索引

select COUNT(*) from 20130107date
where idx_date_long BETWEEN UNIX_TIMESTAMP(&＃39;20110101&＃39;) and UNIX_TIMESTAMP(&＃39;20110102&＃39;);

select COUNT(*) from 20130107date
where idx_date BETWEEN &＃39;20110101&＃39; and &＃39;20110102&＃39;

毫无悬念,两个基本都是瞬时的.