找到 mysql 数据库中的不良索引

为了演示，首先建两个包含不良索引的表，并弄点数据。

mysql> show create table test1/G
*************************** 1. row ***************************
       Table: test1
Create Table: CREATE TABLE `test1` (
  `id` int(11) NOT NULL,
  `f1` int(11) DEFAULT NULL,
  `f2` int(11) DEFAULT NULL,
  `f3` int(11) DEFAULT NULL,
  PRIMARY KEY (`id`),
  KEY `k1` (`f1`,`id`),
  KEY `k2` (`id`,`f1`),
  KEY `k3` (`f1`),
  KEY `k4` (`f1`,`f3`),
  KEY `k5` (`f1`,`f3`,`f2`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1
1 row in set (0.00 sec)

mysql> show create table test2/G
*************************** 1. row ***************************
       Table: test2
Create Table: CREATE TABLE `test2` (
  `id1` int(11) NOT NULL DEFAULT '0',
  `id2` int(11) NOT NULL DEFAULT '0',
  `b` int(11) DEFAULT NULL,
  PRIMARY KEY (`id1`,`id2`),
  KEY `k1` (`b`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1
1 row in set (0.00 sec)

mysql> select count(*) from test2 group by b;                                                                                                        
+----------+
| count(*) |
+----------+
|       32 |
|       17 |
+----------+
2 rows in set (0.00 sec)

1. 包含主键的索引

innodb 本身是聚簇表，每个二级索引本身就包含主键，类似 f1, id 的索引实际虽然没有害处，但反映了使用者对 mysql 索引不了解。而类似 id, f1 的是多余索引，会浪费存储空间，并影响数据更新性能。包含主键的索引用这样一句 sql 就能全部找出来。

mysql> select c.*, pk from 
    ->   (select table_schema, table_name, index_name, concat('|', group_concat(column_name order by seq_in_index separator '|'), '|') cols 
    ->     from INFORMATION_SCHEMA.STATISTICS 
    ->     where index_name != 'PRIMARY' and table_schema != 'mysql'
    -> group by table_schema, table_name, index_name) c,
    ->   (select table_schema, table_name, concat('|', group_concat(column_name order by seq_in_index separator '|'), '|') pk 
    ->     from INFORMATION_SCHEMA.STATISTICS 
    ->     where index_name = 'PRIMARY' and table_schema != 'mysql'
    -> group by table_schema, table_name) p  
    -> where c.table_name = p.table_name and c.table_schema = p.table_schema and c.cols like concat('%', pk, '%');
+--------------+------------+------------+---------+------+
| table_schema | table_name | index_name | cols    | pk   |
+--------------+------------+------------+---------+------+
| test         | test1      | k1         | |f1|id| | |id| |
| test         | test1      | k2         | |id|f1| | |id| |
+--------------+------------+------------+---------+------+
2 rows in set (0.04 sec)

2. 重复索引前缀

包含重复前缀的索引，索引能由另一个包含该前缀的索引完全代替，是多余索引。多余的索引会浪费存储空间，并影响数据更新性能。这样的索引同样用一句 sql 可以找出来。

mysql> select c1.table_schema, c1.table_name, c1.index_name,c1.cols,c2.index_name, c2.cols from
    ->   (select table_schema, table_name, index_name, concat('|', group_concat(column_name order by seq_in_index separator '|'), '|') cols 
    ->     from INFORMATION_SCHEMA.STATISTICS 
    ->     where table_schema != 'mysql' and index_name!='PRIMARY'
    -> group by table_schema,table_name,index_name) c1,   
    ->   (select table_schema, table_name,index_name, concat('|', group_concat(column_name order by seq_in_index separator '|'), '|') cols 
    ->     from INFORMATION_SCHEMA.STATISTICS 
    ->     where table_schema != 'mysql' and index_name != 'PRIMARY'
    -> group by table_schema, table_name, index_name) c2 
    -> where c1.table_name = c2.table_name and c1.table_schema = c2.table_schema and c1.cols like concat(c2.cols, '%') and c1.index_name != c2.index_name;
+--------------+------------+------------+------------+------------+---------+
| table_schema | table_name | index_name | cols       | index_name | cols    |
+--------------+------------+------------+------------+------------+---------+
| test         | test1      | k1         | |f1|id|    | k3         | |f1|    |
| test         | test1      | k4         | |f1|f3|    | k3         | |f1|    |
| test         | test1      | k5         | |f1|f3|f2| | k3         | |f1|    |
| test         | test1      | k5         | |f1|f3|f2| | k4         | |f1|f3| |
+--------------+------------+------------+------------+------------+---------+
4 rows in set (0.02 sec)

3. 低区分度索引

这样的索引由于仍然会扫描大量记录，在实际查询时通常会被忽略。但是在某些情况下仍然是有用的。因此需要根据实际情况进一步分析。这里是区分度小于 10% 的索引，可以根据需要调整参数。

mysql> select p.table_schema, p.table_name, c.index_name, c.car, p.car total from
    ->   (select table_schema, table_name, index_name, max(cardinality) car
    ->     from INFORMATION_SCHEMA.STATISTICS
    -> where index_name != 'PRIMARY'
    -> group by table_schema, table_name,index_name) c,
    ->   (select table_schema, table_name, max(cardinality) car
    ->     from INFORMATION_SCHEMA.STATISTICS
    -> where index_name = 'PRIMARY' and table_schema != 'mysql'
    -> group by table_schema,table_name) p
    -> where c.table_name = p.table_name and c.table_schema = p.table_schema and p.car > 0 and c.car / p.car < 0.1;
+--------------+------------+------------+------+-------+
| table_schema | table_name | index_name | car  | total |
+--------------+------------+------------+------+-------+
| test         | test2      | k1         |    4 |    49 |
+--------------+------------+------------+------+-------+
1 row in set (0.04 sec)

4. 复合主键

由于 innodb 是聚簇表，每个二级索引都会包含主键值。复合主键会造成二级索引庞大，而影响二级索引查询性能，并影响更新性能。同样需要根据实际情况进一步分析。

mysql> select table_schema, table_name, group_concat(column_name order by seq_in_index separator ',') cols, max(seq_in_index) len
    ->    from INFORMATION_SCHEMA.STATISTICS
    ->    where index_name = 'PRIMARY' and table_schema != 'mysql'
    ->    group by table_schema, table_name having len>1;
+--------------+------------+-----------------------------------+------+
| table_schema | table_name | cols                              | len  |
+--------------+------------+-----------------------------------+------+
| test         | test2      | id1,id2                           |    2 |
+--------------+------------+-----------------------------------+------+
1 rows in set (0.01 sec)

（题图来自：webfish.se）