mysql> explain select * from one where last_login='1338252525';
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows| Extra |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
| 1 | SIMPLE | one | ALL | NULL | NULL | NULL | NULL | 20146 | Using where |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
1 row in set (0.00 sec)
以上的两条语句都不是以username开始，这样是用不了索引，通过type=all（全表扫描），key_len=null，rows都很大20146
Ps：one表里只有20003条数据，为什么出现20146，这是优化器对表的一个估算值，不精确的。
Where 子句表达式虽然顺序是(username,password, last_login)或(username,password)但第一个是有范围’<’、’>’，’<=’， ’>=’等出现
mysql> explain select * from one where username>'abgvwfnt' and password ='123456'and last_login='1338251170';
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows| Extra |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
| 1 | SIMPLE | one | ALL | username | NULL | NULL | NULL | 20146 | Using where |
+----+-------------+-------+------+---------------+------+---------+------+-------+-------------+
1 row in set (0.00 sec)
这个查询很明显是遍历所有表，一个索引都没用到，非第一列出现范围（password列或last_login列），则能利用索引到首先出 现范围的一列，也就是“where username='abgvwfnt' and password >'123456'and last_login='1338251170';”或 “where username='abgvwfnt' and password >'123456'and last_login<'1338251170';”索引长度ref_len=43,索引检索到 password列，所以考虑多列索引的时候把那些查询语句用的比较的列放在最后（或非第一位）。
断层，即是where顺序(username, last_login)
mysql> explain select * from one where username='abgvwfnt' and last_login='1338252525';
+----+-------------+-------+------+---------------+----------+---------+-------+------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+------+---------------+----------+---------+-------+------+-------------+
| 1 | SIMPLE | one | ref | username | username | 24 | const |5 | Using where |
+----+-------------+-------+------+---------------+----------+---------+-------+------+-------------+
1 row in set (0.00 sec)
注意这里的key_len=24=8*3(8是username的长度，3是utf8编码)，rows=5，和下面一条sql语句搜索出来一样
mysql> select * from one where username='abgvwfnt';
+-------+----------+----------+-------+------------+
| id | username | password | level | last_login |
+-------+----------+----------+-------+------------+
| 3597 | abgvwfnt | 234567 | 0 | 1338251420 |
| 7693 | abgvwfnt | 456789 | 0 | 1338251717 |
| 11789 | abgvwfnt | 456789 | 0 | 1338251992 |
| 15885 | abgvwfnt | 456789 | 0 | 1338252258 |
| 19981 | abgvwfnt | 456789 | 0 | 1338252525 |
+-------+----------+----------+-------+------------+
5 rows in set (0.00 sec)
mysql> select * from one where username='abgvwfnt' and last_login='1338252525';
+-------+----------+----------+-------+------------+
| id | username | password | level | last_login |
+-------+----------+----------+-------+------------+
| 19981 | abgvwfnt | 456789 | 0 | 1338252525 |
+-------+----------+----------+-------+------------+
1 row in set (0.00 sec)
这个就是要的返回结果，所以可以知道断层(username,last_login)，这样只用到username索引，把用到索引的数据再重新检查 last_login条件，这个相对全表查询来说还是有性能上优化，这也是很多sql优化文章中提到的where 范围查询要放在最后 （这不绝对，但可以利用一部分索引）
如果一个查询whe