物理查询优化之索引

索引是建立在表上的，本质上是通过索引直接定位表的物理元组，加速数据获取的方式，所以索引优化应该归属到物理查询优化阶段。

1. 如何利用索引

通常查询优化器所使用索引的原则如下：

● 索隐裂座位条件出现在WHERE、H***ING、ON 子句中，这样有利于索引过滤元组；

● 索引列是被链接的表对象的列且存在于连接条件中；

● 还有一些情况可以使用索引，如排序操作、在索引列上球MIN、MAX等。

1.1 示例：

create table A(a1 int unique,a2 varchar(10),a3 int); （a1列上创建隐含索引）

● 对表做查询，没有列对象作为过滤条件（如出现在WHERE子句中），只能顺序扫描

mysql> explain extended select * from A;

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------+

| 1 | SIMPLE | A | ALL | NULL | NULL | NULL | NULL | 1 | 100.00 | NULL |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------+

type = ALL 表示 顺序扫描或范围扫描，不是用索引，顺序扫描，直接读取表上的数据（访问数据文件）。

● 对表做查询，有列对象列且索引列作为过滤条件，可作为索引扫描

mysql> explain extended select * from A where a1 >2;

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

| 1 | SIMPLE | A | range | a1 | a1 | 5 | NULL | 1 | 100.00 | Using index condition |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

t ype = range 表示基于索引做范围扫描。Using index condition 表示尝试通过索引来获取

● 对表做查询，索引列被运算符“-” 处理，查询优化器不能再执行前进行取反操作，不可利用索引，只能顺序扫描。

select A.* from A where -A.a1 = -2; (不能利用索引)

select A.* from A where A.a1 = 2; (利用索引)

●对表做查询，选择条件不包括索引列，只能顺序扫描

select A.* from A where A.a3 = 2;

● 索引列参与了运算，不能使用索引

select A.*from A whereA.a1 + A.a3 = 2;

select A.*from A whereA.a1= A.a3 + 2;

select A.*from A whereA.a1< 1+2;（可以使用索引）

● 查询时，索引列对象座位过滤条件时，操作符是 <> 时，不可做索引扫描，如果是 < 或者 > 时，可以使用索引

mysql> explain extended select A.* from A where A.a1 <> 2;(不使用索引)

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------------+

| 1 | SIMPLE | A | ALL | a1 | NULL | NULL | NULL | 2 | 100.00 | Using where |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------------+

mysql> explain extended select A.* from A where A.a1 > 2;(使用索引)

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

| 1 | SIMPLE | A | range | a1 | a1 | 5 | NULL | 1 | 100.00 | Using index condition |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

● 查询时，索引列对象座位过滤条件时，操作符是 between - and 时，可以使用索引

mysql> explain extended select A.* from A where A.a1 between 1 and 2;

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

| 1 | SIMPLE | A | range | a1 | a1 | 5 | NULL | 2 | 100.00 | Using index condition |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-----------------------+

1.2 所以，对于索引列，索引可用的条件如下：

● 在WHERE, JOIN/ON/H***ING 的条件中出现 “key <op> 常量” 格式的条件子句（索引列不能参与带有变量表达式的运算）

● 操作符不能是<>操作符（不等于操作符在任何类型的列上都不能使用索引）

● 索引列的选择率越低，索引越有效，通常认为选择率小鱼0.1的索引扫描效果会好一些

2. 索引列的位置对使用索引的影响

2.1 对目标列、WHERE 等条件子句的影响

● 索引列出现在目标列通常不可以使用索引（但不是全部情况），索引列出现目标列对查询语句的优化没有好的影响（PostgreSQL），如：

select A.1a from A;（（PostgreSQL））

注意：具体的数据库可能存在差别，在我的mysql环境是可以用到索引的：如下：

mysql> explain extended select A.a1 from A;

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-------------+

| 1 | SIMPLE | A | index | NULL | a1 | 5 | NULL | 2 | 100.00 | Using index |

+----+-------------+-------+-------+---------------+------+---------+------+------+----------+-------------+

● 聚集函数MIN/MAX 用在索引列上，出现在目标列中，可以使用索引（PostgreSQL）。如：

select MAX(A.a1) from A;（PostgreSQL）

注意：具体的数据库可能存在差别，在我的mysql环境没有用到索引：如下：

mysql> explain extended select Max(A.a1) from A;

+----+-------------+-------+------+---------------+------+---------+------+------+----------+------------------------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+------------------------------+

| 1 | SIMPLE | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | Select tables optimized away |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+------------------------------+()

● 索引出现在JOIN/ON 子句中，作为连接条件，不可使用索引。但是索引列出现限制条件满足“key<op>常量”格式可用到索引。

create table B(b1 int unique,b2 varchar(10),b3 int);

mysql> explain extended select A.*,B.* from A join B on a1 = b1;

+----+-------------+-------+------+---------------+------+---------+-----------+------+----------+-------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+------+---------+-----------+------+----------+-------------+

| 1 | SIMPLE | B | ALL | b1 | NULL | NULL | NULL | 1 | 100.00 | Using where |

| 1 | SIMPLE | A | ref | a1 | a1 | 5 | test.B.b1 | 1 | 100.00 | NULL |

+----+-------------+-------+------+---------------+------+---------+-----------+------+----------+-------------+

执行计划表示 对B表进行了全表扫描，对A表连接时使用了索引。

mysql> explain extended select A.*,B.* from A join B on a1 = b1 and a1 = 1;

+----+-------------+-------+-------+---------------+------+---------+-------+------+----------+-------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+-------+---------------+------+---------+-------+------+----------+-------+

| 1 | SIMPLE | A | const | a1 | a1 | 5 | const | 1 | 100.00 | NULL |

| 1 | SIMPLE | B | const | b1 | b1 | 5 | const | 1 | 100.00 | NULL |

+----+-------------+-------+-------+---------------+------+---------+-------+------+----------+-------+

对A和B都使用而索引。

● 还有些关于order by,group by 的，发现数据库不同的话存在差异，就不写了，

当然上面的也可能存在差异，具体的话在用的时候看下执行计划就知道了。

3. 联合索引对索引使用的影响

create table E(e1 int,e2 varchar(10),e3 int,e4 int,primary key(e1,e3,e4));

使用联合索引的前缀部分索引键，可触发索引的使用；

使用部分索引键，但不是联合索引的前缀部分，不可出发索引的使用。

select * from E where e1 = 1;

select * from E where e1 = 1 and e3 =1 ;（select * from E where e1 = 1 or e3 =1 ;这种情况不会用到索引）

select * from E where e1 = 1 and e3 =1 and e4 = 1;

以上三个都是可以使用索引的：

mysql> explain extended select * from E where e1 = 1;

+----+-------------+-------+------+---------------+---------+---------+-------+------+----------+-------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+---------+---------+-------+------+----------+-------+

| 1 | SIMPLE | E | ref | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL |

+----+-------------+-------+------+---------------+---------+---------+-------+------+----------+-------+

1 row in set, 1 warning (0.00 sec)

mysql> explain extended select * from E where e1 = 1 and e3 =1;

+----+-------------+-------+------+---------------+---------+---------+-------------+------+----------+-------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+---------+---------+-------------+------+----------+-------+

| 1 | SIMPLE | E | ref | PRIMARY | PRIMARY | 8 | const,const | 1 | 100.00 | NULL |

+----+-------------+-------+------+---------------+---------+---------+-------------+------+----------+-------+

1 row in set, 1 warning (0.00 sec)

mysql> explain extended select * from E where e1 = 1 and e3 =1 and e4 = 1;

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-----------------------------------------------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-----------------------------------------------------+

| 1 | SIMPLE | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | Impossible WHERE noticed after reading const tables |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-----------------------------------------------------+

注意：这里第三个看上去没有使用到索引，那是因为 e1 = 1 and e3 =1这两个条件过滤后就没有数据了，后面的e4=1不用看也知道返回false了。

下面的这种是不会用到索引的：

mysql> explain extended select * from E where e3 = 1 and e4 = 1;

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------------+

| 1 | SIMPLE | E | ALL | NULL | NULL | NULL | NULL | 1 | 100.00 | Using where |

+----+-------------+-------+------+---------------+------+---------+------+------+----------+-------------+

另外这种情况：虽然使用到了索引，但是只是使用到了e1这一列上的索引，e4=1 这里没有用到索引。

mysql> explain extended select * from E where e1 = 1 and e4 = 1;

+----+-------------+-------+------+---------------+---------+---------+-------+------+----------+-------------+

| id | select_type | table | type | possible_keys | key | key_len | ref | rows | filtered | Extra |

+----+-------------+-------+------+---------------+---------+---------+-------+------+----------+-------------+

| 1 | SIMPLE | E | ref | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | Using where |

+----+-------------+-------+------+---------------+---------+---------+-------+------+----------+-------------+

4.多个索引对索引使用的影响