时间:2021-07-01 10:21:17 帮助过:4人阅读
从结果中可以到titles表的主索引为<emp_no, title, from_date>,还有一个辅助索引<emp_no>。为了避免多个索引使事情变复杂(MySQL的SQL优化器在多索引时行为比较复杂),这里我们将辅助索引drop掉:ALTER TABLE employees.titles DROP INDEX emp_no;这样就可以专心分析索引PRIMARY的行为了。
1 EXPLAIN SELECT * FROM employees.titles WHERE emp_no=‘10001‘ AND title=‘Senior Engineer‘ AND from_date=‘1986-06-26‘; 2 +----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+ 5 | 1 | SIMPLE | titles | const | PRIMARY | PRIMARY | 59 | const,const,const | 1 | | 6 +----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+
很明显,当按照索引中所有列进行精确匹配(这里精确匹配指“=”或“IN”匹配)时,索引可以被用到。这里有一点需要注意,理论上索引对顺序是敏感的,但是由于MySQL的查询优化器会自动调整where子句的条件顺序以使用适合的索引,例如我们将where中的条件顺序颠倒:
1 EXPLAIN SELECT * FROM employees.titles WHERE from_date=‘1986-06-26‘ AND emp_no=‘10001‘ AND title=‘Senior Engineer‘; 2 +----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+ 5 | 1 | SIMPLE | titles | const | PRIMARY | PRIMARY | 59 | const,const,const | 1 | | 6 +----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+
效果是一样的。
1 EXPLAIN SELECT * FROM employees.titles WHERE emp_no=‘10001‘; 2 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------+ 5 | 1 | SIMPLE | titles | ref | PRIMARY | PRIMARY | 4 | const | 1 | | 6 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------+
当查询条件精确匹配索引的左边连续一个或几个列时,如<emp_no>或<emp_no, title>,所以可以被用到,但是只能用到一部分,即条件所组成的最左前缀。上面的查询从分析结果看用到了PRIMARY索引,但是key_len为4,说明只用到了索引的第一列前缀。
1 EXPLAIN SELECT * FROM employees.titles WHERE emp_no=‘10001‘ AND from_date=‘1986-06-26‘; 2 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+ 5 | 1 | SIMPLE | titles | ref | PRIMARY | PRIMARY | 4 | const | 1 | Using where | 6 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+
此时索引使用情况和情况二相同,因为title未提供,所以查询只用到了索引的第一列,而后面的from_date虽然也在索引中,但是由于title不存在而无法和左前缀连接,因此需要对结果进行扫描过滤from_date(这里由于emp_no唯一,所以不存在扫描)。如果想让from_date也使用索引而不是where过滤,可以增加一个辅助索引<emp_no, from_date>,此时上面的查询会使用这个索引。除此之外,还可以使用一种称之为“隔离列”的优化方法,将emp_no与from_date之间的“坑”填上。
首先我们看下title一共有几种不同的值:
1 SELECT DISTINCT(title) FROM employees.titles; 2 +--------------------+ 3 | title | 4 +--------------------+ 5 | Senior Engineer | 6 | Staff | 7 | Engineer | 8 | Senior Staff | 9 | Assistant Engineer | 10 | Technique Leader | 11 | Manager | 12 +--------------------+
只有7种。在这种成为“坑”的列值比较少的情况下,可以考虑用“IN”来填补这个“坑”从而形成最左前缀:
1 EXPLAIN SELECT * FROM employees.titles 2 WHERE emp_no=‘10001‘ 3 AND title IN (‘Senior Engineer‘, ‘Staff‘, ‘Engineer‘, ‘Senior Staff‘, ‘Assistant Engineer‘, ‘Technique Leader‘, ‘Manager‘) 4 AND from_date=‘1986-06-26‘; 5 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 6 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 7 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 8 | 1 | SIMPLE | titles | range | PRIMARY | PRIMARY | 59 | NULL | 7 | Using where | 9 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
这次key_len为59,说明索引被用全了,但是从type和rows看出IN实际上执行了一个range查询,这里检查了7个key。
看下两种查询的性能比较:
1 SHOW PROFILES; 2 +----------+------------+-------------------------------------------------------------------------------+ 3 | Query_ID | Duration | Query | 4 +----------+------------+-------------------------------------------------------------------------------+ 5 | 10 | 0.00058000 | SELECT * FROM employees.titles WHERE emp_no=‘10001‘ AND from_date=‘1986-06-26‘| 6 | 11 | 0.00052500 | SELECT * FROM employees.titles WHERE emp_no=‘10001‘ AND title IN ... | 7 +----------+------------+-------------------------------------------------------------------------------+
“填坑”后性能提升了一点。如果经过emp_no筛选后余下很多数据,则后者性能优势会更加明显。当然,如果title的值很多,用填坑就不合适了,必须建立辅助索引。
1 EXPLAIN SELECT * FROM employees.titles WHERE from_date=‘1986-06-26‘; 2 +----+-------------+--------+------+---------------+------+---------+------+--------+-------------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+------+---------------+------+---------+------+--------+-------------+ 5 | 1 | SIMPLE | titles | ALL | NULL | NULL | NULL | NULL | 443308 | Using where | 6 +----+-------------+--------+------+---------------+------+---------+------+--------+-------------+
由于不是最左前缀,索引这样的查询显然用不到索引。
此时可以用到索引,但是如果通配符不是只出现在末尾,则无法使用索引。
1 EXPLAIN SELECT * FROM employees.titles WHERE emp_no=‘10001‘ AND title LIKE ‘Senior%‘; 2 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 5 | 1 | SIMPLE | titles | range | PRIMARY | PRIMARY | 56 | NULL | 1 | Using where | 6 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
1 EXPLAIN SELECT * FROM employees.titles WHERE emp_no<‘10010‘ and title=‘Senior Engineer‘; 2 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 5 | 1 | SIMPLE | titles | range | PRIMARY | PRIMARY | 4 | NULL | 16 | Using where | 6 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
范围列可以用到索引(必须是最左前缀),但是范围列后面的列无法用到索引。同时,索引最多用于一个范围列,因此如果查询条件中有两个范围列则无法全用到索引。
1 EXPLAIN SELECT * FROM employees.titles 2 WHERE emp_no<‘10010‘ 3 AND title=‘Senior Engineer‘ 4 AND from_date BETWEEN ‘1986-01-01‘ AND ‘1986-12-31‘; 5 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 6 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 7 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 8 | 1 | SIMPLE | titles | range | PRIMARY | PRIMARY | 4 | NULL | 16 | Using where | 9 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
可以看到索引对第二个范围索引无能为力。这里特别要说明MySQL一个有意思的地方,那就是仅用explain可能无法区分范围索引和多值匹配,因为在type中这两者都显示为range。同时,用了“between”并不意味着就是范围查询,例如下面的查询:
1 EXPLAIN SELECT * FROM employees.titles 2 WHERE emp_no BETWEEN ‘10001‘ AND ‘10010‘ 3 AND title=‘Senior Engineer‘ 4 AND from_date BETWEEN ‘1986-01-01‘ AND ‘1986-12-31‘; 5 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 6 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 7 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ 8 | 1 | SIMPLE | titles | range | PRIMARY | PRIMARY | 59 | NULL | 16 | Using where | 9 +----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
看起来是用了两个范围查询,但作用于emp_no上的“BETWEEN”实际上相当于“IN”,也就是说emp_no实际是多值精确匹配。可以看到这个查询用到了索引全部三个列。因此在MySQL中要谨慎地区分多值匹配和范围匹配,否则会对MySQL的行为产生困惑。
很不幸,如果查询条件中含有函数或表达式,则MySQL不会为这列使用索引(虽然某些在数学意义上可以使用)。例如:
1 EXPLAIN SELECT * FROM employees.titles WHERE emp_no=‘10001‘ AND left(title, 6)=‘Senior‘; 2 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+ 5 | 1 | SIMPLE | titles | ref | PRIMARY | PRIMARY | 4 | const | 1 | Using where | 6 +----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+
虽然这个查询和情况五中功能相同,但是由于使用了函数left,则无法为title列应用索引,而情况五中用LIKE则可以。再如:
EXPLAIN SELECT * FROM employees.titles WHERE emp_no - 1=‘10000‘; +----+-------------+--------+------+---------------+------+---------+------+--------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+--------+------+---------------+------+---------+------+--------+-------------+ | 1 | SIMPLE | titles | ALL | NULL | NULL | NULL | NULL | 443308 | Using where | +----+-------------+--------+------+---------------+------+---------+------+--------+-------------+
显然这个查询等价于查询emp_no为10001的函数,但是由于查询条件是一个表达式,MySQL无法为其使用索引。看来MySQL还没有智能到自动优化常量表达式的程度,因此在写查询语句时尽量避免表达式出现在查询中,而是先手工私下代数运算,转换为无表达式的查询语句。
既然索引可以加快查询速度,那么是不是只要是查询语句需要,就建上索引?答案是否定的。因为索引虽然加快了查询速度,但索引也是有代价的:索引文件本身要消耗存储空间,同时索引会加重插入、删除和修改记录时的负担,另外,MySQL在运行时也要消耗资源维护索引,因此索引并不是越多越好。一般两种情况下不建议建索引。
第一种情况是表记录比较少,例如一两千条甚至只有几百条记录的表,没必要建索引,让查询做全表扫描就好了。至于多少条记录才算多,这个个人有个人的看法,我个人的经验是以2000作为分界线,记录数不超过 2000可以考虑不建索引,超过2000条可以酌情考虑索引。
另一种不建议建索引的情况是索引的选择性较低。所谓索引的选择性(Selectivity),是指不重复的索引值(也叫基数,Cardinality)与表记录数(#T)的比值:
Index Selectivity = Cardinality / #T
显然选择性的取值范围为(0, 1],选择性越高的索引价值越大,这是由B+Tree的性质决定的。例如,上文用到的employees.titles表,如果title字段经常被单独查询,是否需要建索引,我们看一下它的选择性:
1 SELECT count(DISTINCT(title))/count(*) AS Selectivity FROM employees.titles; 2 +-------------+ 3 | Selectivity | 4 +-------------+ 5 | 0.0000 | 6 +-------------+
title的选择性不足0.0001(精确值为0.00001579),所以实在没有什么必要为其单独建索引。
有一种与索引选择性有关的索引优化策略叫做前缀索引,就是用列的前缀代替整个列作为索引key,当前缀长度合适时,可以做到既使得前缀索引的选择性接近全列索引,同时因为索引key变短而减少了索引文件的大小和维护开销。下面以employees.employees表为例介绍前缀索引的选择和使用。
从图12可以看到employees表只有一个索引<emp_no>,那么如果我们想按名字搜索一个人,就只能全表扫描了:
1 EXPLAIN SELECT * FROM employees.employees WHERE first_name=‘Eric‘ AND last_name=‘Anido‘; 2 +----+-------------+-----------+------+---------------+------+---------+------+--------+-------------+ 3 | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | 4 +----+-------------+-----------+------+---------------+------+---------+------+--------+-------------+ 5 | 1 | SIMPLE | employees | ALL | NULL | NULL | NULL | NULL | 300024 | Using where | 6 +----+-------------+-----------+------+---------------+------+---------+------+--------+-------------+
如果频繁按名字搜索员工,这样显然效率很低,因此我们可以考虑建索引。有两种选择,建<first_name>或<first_name, last_name>,看下两个索引的选择性:
1 SELECT count(DISTINCT(first_name))/count(*) AS Selectivity FROM employees.employees; 2 +-------------+ 3 | Selectivity | 4 +-------------+ 5 | 0.0042 | 6 +-------------+ 7 8 SELECT count(DISTINCT(concat(first_name, last_name)))/count(*) AS Selectivity FROM employees.employees; 9 +-------------+ 10 | Selectivity | 11 +-------------+ 12 | 0.9313 | 13 +-------------+
<first_name>显然选择性太低,<first_name, last_name>选择性很好,但是first_name和last_name加起来长度为30,有没有兼顾长度和选择性的办法?可以考虑用first_name和last_name的前几个字符建立索引,例如<first_name, left(last_name, 3)>,看看其选择性:
1 SELECT count(DISTINCT(concat(first_name, left(last_name, 3))))/count(*) AS Selectivity FROM employees.employees; 2 +-------------+ 3 | Selectivity | 4 +-------------+ 5 | 0.7879 | 6 +-------------+ 7 选择性还不错,但离0.9313还是有点距离,那么把last_name前缀加到4: 8 9 SELECT count(DISTINCT(concat(first_name, left(last_name, 4))))/count(*) AS Selectivity FROM employees.employees; 10 +-------------+ 11 | Selectivity | 12 +-------------+ 13 | 0.9007 | 14 +-------------+ 15 这时选择性已经很理想了,而这个索引的长度只有18,比<first_name, last_name>短了接近一半,我们把这个前缀索引 建上: 16 17 ALTER TABLE employees.employees 18 ADD INDEX `first_name_last_name4` (first_name, last_name(4)); 19 此时再执行一遍按名字查询,比较分析一下与建索引前的结果: 20 21 SHOW PROFILES; 22 +----------+------------+---------------------------------------------------------------------------------+ 23 | Query_ID | Duration | Query | 24 +----------+------------+---------------------------------------------------------------------------------+ 25 | 87 | 0.11941700 | SELECT * FROM employees.employees WHERE first_name=‘Eric‘ AND last_name=‘Anido‘ | 26 | 90 | 0.00092400 | SELECT