MySQL单列索引和联合索引

本文通过一个案例，介绍优化器对单列索引和联合索引的选择。

order表的ord_seq字段上有2个索引，单列索引( order_seq)和联合索引(order_seq,order_type)

MySQL > explain select * from `order` where order_seq = 1502131212577;
+----+-------------+-------+------------+------+-------------------------------+------------+---------+-------+------+----------+-------+
| id | select_type | table | partitions | type | possible_keys                 | key        | key_len | ref   | rows | filtered | Extra |
+----+-------------+-------+------------+------+-------------------------------+------------+---------+-------+------+----------+-------+
|  1 | SIMPLE      | order | NULL       | ref  | idx_ordseq,idx_ordseq_ordtype | idx_ordseq | 8       | const |    1 |   100.00 | NULL  |
+----+-------------+-------+------------+------+-------------------------------+------------+---------+-------+------+----------+-------+

可以看到优化器选择了单列索引idx_ordseq( order_seq)，而不是联合索引。因为该索引的叶子节点包含单个键值，所以理论上一个页可以存放更多的记录。

如果换种场景：

MySQL > explain select * from `order` where order_seq = 1502131212577 order by order_type desc;
+----+-------------+-------+------------+------+-------------------------------+--------------------+---------+-------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys                 | key                | key_len | ref   | rows | filtered | Extra       |
+----+-------------+-------+------------+------+-------------------------------+--------------------+---------+-------+------+----------+-------------+
|  1 | SIMPLE      | order | NULL       | ref  | idx_ordseq,idx_ordseq_ordtype | idx_ordseq_ordtype | 8       | const |    1 |   100.00 | Using where |
+----+-------------+-------+------------+------+-------------------------------+--------------------+---------+-------+------+----------+-------------+
1 row in set, 1 warning (0.00 sec)

where条件字段和排序字段添加联合索引，解决了filesort的问题。