mysql索引之前缀索引
有时候需要索引很长的字符列,这会让索引变得大且慢。通常可以索引开始的部分字符,这样可以大大节约索引空间,从而提高索引效率。但这样也会降低索引的选择性。索引的选择性是指不重复的索引值(也称为基数,cardinality)和数据表的记录总数的比值,范围从1/T到1之间。索引的选择性越高则查询效率越高,因为选择性高的索引可以让MySQL在查找时过滤掉更多的行。唯一索引的选择性是1,这是最好的索引选择性,性能也是最好的。
一般情况下某个前缀的选择性也是足够高的,足以满足查询性能。对于BLOB,TEXT,或者很长的VARCHAR类型的列,必须使用前缀索引,因为MySQL不允许索引这些列的完整长度。诀窍在于要选择足够长的前缀以保证较高的选择性,同时又不能太长(以便节约空间)。前缀应该足够长,以使得前缀索引的选择性接近于索引的整个列。换句话说,前缀的”基数“应该接近于完整的列的”基数“。
为了决定前缀的合适长度,需要找到最常见的值的列表,然后和最常见的前缀列表进行比较。下面的示例是mysql官方提供的示例数据库,下载地址如下:
http://downloads.mysql.com/docs/sakila-db.zip
在示例数据库sakila中并没有合适的例子,所以从表city中生成一个示例表,这样就有足够数据进行演示:
1.解压下载的sakila-db.zip文件
2.使用source命令以及sakila-schema.sql和sakila-data.sql文件来初始化sakila库以及相关表格
mysql> select database();
+------------+
| database() |
+------------+
| sakila |
+------------+
1 row in set (0.00 sec)
mysql> create table city_demo (city varchar(50) not null);
mysql> insert into city_demo (city) select city from city; --执行两次
Query OK, 600 rows affected (0.02 sec)
Records: 600 Duplicates: 0 Warnings: 0
mysql> update city_demo set city = ( select city from city order by rand() limit 1);
Query OK, 1198 rows affected (0.42 sec)
Rows matched: 1200 Changed: 1198 Warnings: 0
注:因为上述sql语句使用了rand函数,所以每个人的执行结果可以都不一样!
首先找到最常见的城市列表:
mysql> select count(*) as cnt,city from city_demo group by city order by cnt desc limit 10;
+-----+---------------+
| cnt | city |
+-----+---------------+
| 8 | Dongying |
| 7 | Omdurman |
| 6 | Etawah |
| 6 | Okara |
| 6 | Tsuyama |
| 6 | Brindisi |
| 6 | Kuwana |
| 6 | Grand Prairie |
| 5 | Fuyu |
| 5 | Siegen |
+-----+---------------+
10 rows in set (0.00 sec)
现在查找到频繁出现的城市前缀。先从3个前缀字母开始,然后4个,5个,6个:
mysql> select count(*) as cnt,left(city,3) as pref from city_demo group by pref order by cnt desc limit 10;
+-----+------+
| cnt | pref |
+-----+------+
| 23 | San |
| 15 | Hal |
| 14 | Cha |
| 14 | al- |
| 12 | Bat |
| 12 | Kor |
| 11 | Don |
| 11 | Shi |
| 10 | La |
| 10 | El |
+-----+------+
10 rows in set (0.00 sec)
可以看到3字节检索到的结果与全文检索相差很大,继续增加到4个字节
mysql> select count(*) as cnt,left(city,4) as pref from city_demo group by pref order by cnt desc limit 10;
+-----+------+
| cnt | pref |
+-----+------+
| 14 | San |
| 8 | Dong |
| 7 | Iwak |
| 7 | al-Q |
| 7 | Omdu |
| 6 | Kuwa |
| 6 | Tsuy |
| 6 | Brin |
| 6 | Etaw |
| 6 | Okar |
+-----+------+
10 rows in set (0.00 sec)
mysql> select count(*) as cnt,left(city,5) as pref from city_demo group by pref order by cnt desc limit 10;
+-----+-------+
| cnt | pref |
+-----+-------+
| 8 | Dongy |
| 7 | al-Qa |
| 7 | Omdur |
| 6 | Okara |
| 6 | Valle |
| 6 | Grand |
| 6 | Tsuya |
| 6 | Etawa |
| 6 | South |
| 6 | Kuwan |
+-----+-------+
10 rows in set (0.00 sec)
mysql> select count(*) as cnt,left(city,6) as pref from city_demo group by pref order by cnt desc limit 10;
+-----+--------+
| cnt | pref |
+-----+--------+
| 8 | Dongyi |
| 7 | Omdurm |
| 6 | Okara |
| 6 | Tsuyam |
| 6 | Valle |
| 6 | Grand |
| 6 | Etawah |
| 6 | Brindi |
| 6 | Kuwana |
| 5 | Haldia |
+-----+--------+
10 rows in set (0.01 sec)
通过上面改变不同前缀长度发现,当前缀长度为6时,这个前缀的选择性就接近完整列的选择性了。
当然还有另外更方便的方法,那就是计算完整列的选择性,并使其前缀的选择性接近于完整列的选择性。下面显示如何计算完整列的选择性:
mysql> select count(distinct city)/count(*) from city_demo;
+-------------------------------+
| count(distinct city)/count(*) |
+-------------------------------+
| 0.4333 |
+-------------------------------+
1 row in set (0.00 sec)
可以在一个查询中针对不同前缀长度的选择性进行计算,这对于大表非常有用,下面给出如何在同一个查询中计算不同前缀长度的选择性:
mysql> select count(distinct left(city,3))/count(*) as sel3,count(distinct left(city,4))/count(*) as sel4,count(distinct left(city,5))/count(*) as sel5, count(distinct left(city,6))/count(*) as sel6 from city_demo;
+--------+--------+--------+--------+
| sel3 | sel4 | sel5 | sel6 |
+--------+--------+--------+--------+
| 0.3408 | 0.4100 | 0.4225 | 0.4300 |
+--------+--------+--------+--------+
1 row in set (0.00 sec)
可以看见当索引前缀为6时的基数是0.4300,已经接近完整列选择性0.4333。
下面根据找到的索引前缀长度创建前缀索引:
mysql> alter table city_demo add key (city(6));
Query OK, 0 rows affected (0.19 sec)
Records: 0 Duplicates: 0 Warnings: 0
mysql> explain select * from city_demo where city like 'Jin%';
*************************** 1. row ***************************
id: 1
select_type: SIMPLE
table: city_demo
partitions: NULL
type: range
possible_keys: city
key: city
key_len: 8
ref: NULL
rows: 4
filtered: 100.00
Extra: Using where
1 row in set, 1 warning (0.00 sec)
可以看见正确使用刚创建的索引。
优点:前缀索引是一种能使索引更小,更快的有效办法
缺点:mysql无法使用其前缀索引做ORDER BY和GROUP BY,也无法使用前缀索引做覆盖扫描。