mysql - mysql中覆盖索引是否比最左前缀索引策略优先级高？

Question

我有一个名为 test 的表:

mysql> describe test;
+-------+---------+------+-----+---------+-------+
| Field | Type    | Null | Key | Default | Extra |
+-------+---------+------+-----+---------+-------+
| coll1 | int(11) | YES  | MUL | NULL    |       |
| coll2 | int(11) | YES  |     | NULL    |       |
| coll3 | int(11) | YES  |     | NULL    |       |
| coll4 | int(11) | YES  |     | NULL    |       |
+-------+---------+------+-----+---------+-------+

我在 table 上有一个索引:

mysql> show index from test;
+-------+------------+----------+--------------+-------------+-----------+-------------+----------+--------+------  +------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+-------+------------+----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| test  |          1 | hi       |            1 | coll1       | A         |           0 |     NULL | NULL   | YES  | BTREE       |         |               |
| test  |          1 | hi       |            2 | coll2       | A         |           0 |     NULL | NULL   | YES  | BTREE       |         |               |
| test  |          1 | hi       |            3 | coll3       | A         |           0 |     NULL | NULL   | YES  | BTREE       |         |               |
+-------+------------+----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+

由于采用最左边前缀索引策略，下面的输出很容易理解。

mysql> explain select * from test where coll2=1;
+----+-------------+-------+------+---------------+------+---------+------+------+-------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra       |
+----+-------------+-------+------+---------------+------+---------+------+------+-------------+
|  1 | SIMPLE      | test  | ALL  | NULL          | NULL | NULL    | NULL |    1 | Using where |
+----+-------------+-------+------+---------------+------+---------+------+------+-------------+

我知道覆盖索引的东西，但是这是怎么发生的呢？这是否意味着覆盖索引的优先级高于最左边的前缀索引策略？

mysql> explain select coll1 from test where coll2=1;
+----+-------------+-------+-------+---------------+------+---------+------+------+--------------------------+
| id | select_type | table | type  | possible_keys | key  | key_len | ref  | rows | Extra                    |
+----+-------------+-------+-------+---------------+------+---------+------+------+--------------------------+
|  1 | SIMPLE      | test  | index | NULL          | hi   | 15      | NULL |    1 | Using where; Using index |
+----+-------------+-------+-------+---------------+------+---------+------+------+--------------------------+

附注 我使用的是mysql版本5.6.13

Answer 1

抱歉我的英语不好。

表文件由DATA文件 + INDEX文件组成。通常，一个复合索引列是所有列的子集。这意味着某些 INDEX 文件大小小于 DATA 文件大小。因此，以下查询使用 hi INDEX 扫描，其读取的磁盘数量少于 DATA 完整扫描。

SELECT coll1 FROM test WHERE coll2=1;

但是当您按如下方式检索 coll1, coll4 时，coll4 不是 hi 索引的一部分。因此，完整数据扫描速度更快。

SELECT coll1, coll4 FROM test WHERE coll2=1;