【转】Mysql索引设计原则

来源：https://segmentfault.com/a/1190000000473085

假设一高频查询如下
SELECT * FROM user WHERE area='amoy' AND sex=0 ORDER BY last_login DESC limit 30;
如何建立索引？描述考虑的过程

user表如下：
初始化100W条数据，其中，area要通过IP查询生成，sex为 0,1 随机

CREATE TABLE user (
id int(10) NOT NULL AUTO_INCREMENT COMMENT '自增编号',
username varchar(30) NOT NULL DEFAULT '0' COMMENT '用户名',
password varchar(30) NOT NULL DEFAULT '0' COMMENT '密码',
area varchar(30) NOT NULL COMMENT '地址',
sex int(10) NOT NULL COMMENT '性别1，男；2，女。',
last_login int(10) NOT NULL COMMENT '最近一次登录时间戳',
PRIMARY KEY (id)
) ENGINE=InnoDB AUTO_INCREMENT=892013 DEFAULT CHARSET=latin1

最终我的索引
(last_login,area)

索引原则：

1.where和order by等的字段建立索引

2.使用唯一索引：对于last_login,area等字段重复的次数比较少，可以使用索引；而sex无非就两个值：性别1，男；2，不值得索引

3.多列索引：不要为每一个列单独建立索引，这样并不能将mysql索引的效率最大化。使用“索引合并策略”

4.选择合理的索引列顺序：索引列的顺序意味着索引首先按照最左列进行排序，然后是第二列，以此类推。如(last_login,area)会先按照 last_login 进行排序，然后才是area。

5.将选择性最高的索引放到前面，也就是会所按照这个条件搜索到的数据最少，选择性就越高，比如选择性：last_login> area> sex。

6.索引不是越多越好，适合的索引可以提高查询效率，但是会降低写入效率，根据项目保持两者的平衡性最好了。

总结上面，首先sex不适合建立索引，有没有索引对于效率的提升意义不大，其次索引会按照最左列进行排序，因此将last_login放到最前面。

测试过程：

user表
没有任何索引的查询相关日志：
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.57s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.56s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.55s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.59s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.55s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.55s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.57s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.58s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.57s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.57s
共花费时间：5.66s

建立索引area:
ALTER TABLE user ADD INDEX index_area (area) ;
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.06s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.10s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.11s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.20s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.07s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
共花费时间：0.66s
可见，建立area以后对性能的影响是巨大的(5.66/0.66 约为8.5758倍)
删除索引：ALTER TABLE user DROP INDEX index_area;
删除area索引发现时间又变成了0.57s

建立last_login索引：
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.03s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.09s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.51s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.07s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.06s
共花费时间：0.87s
同样能够提升性能（5.66/0.87 约为6.5057倍）

建立sex索引：
ALTER TABLE user ADD INDEX index_sex (sex) ;
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.87s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.87s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.87s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.89s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.88s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.87s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.86s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.88s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.87s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.87s
共花费时间：8.73s
同样能够提升性能（5.66s/8.73 约为0.6483倍）效率反而降低了？？求解？
建立这个sex索引还不如不建。

删除索引：
ALTER TABLE user DROP INDEX index_sex;
发现时间又变成了0.57s左右，

建立两个单独的索引：
ALTER TABLE user
ADD INDEX index_area (area) ,
ADD INDEX index_last_login (last_login) ;

SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.09s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.33s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.21s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.28s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.03s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.67s

发现建立两个单独的索引还不如只建立一个索引
删除索引：
发现时间又变成了0.57s左右，

建立一个的联合索引：
ALTER TABLE user
ADD INDEX index_last_login_area (last_login,area) ,
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.00s
额，第二条数据这是怎么了，我测试了5次都在这附近晃悠哈！
这尼玛，找对索引啦！就该这么建立，查询不出来需要的时间啦！估计就是我们需要的索引啦！！！！

删除索引：
发现时间又变成了0.57s左右，

建立一个的联合索引：
ALTER TABLE user
ADD INDEX index_sex_last_login_area (sex,last_login,area)
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.18s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.17s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.81s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.01s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.03s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
sex怎么总是你在拖后腿啊！把你调整到索引的最后一个吧！
删除索引：
发现时间又变成了0.57s左右，

建立一个的联合索引：
ALTER TABLE user
ADD INDEX index_last_login_area_sex (area,last_login,sex)
SELECT * FROM user WHERE area='美国ATT用户' AND sex=0 ORDER BY last_login DESC limit 30; 0.03s
SELECT * FROM user WHERE area='泰国' AND sex=0 ORDER BY last_login DESC limit 30; 0.07s
SELECT * FROM user WHERE area='台湾省台湾大宽频' AND sex=0 ORDER BY last_login DESC limit 30; 0.50s
SELECT * FROM user WHERE area='美国弗吉尼亚州' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='德国奔驰汽车' AND sex=0 ORDER BY last_login DESC limit 30; 0.05s
SELECT * FROM user WHERE area='台湾省中华电信' AND sex=0 ORDER BY last_login DESC limit 30; 0.06s
SELECT * FROM user WHERE area='韩国' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='拉美地区' AND sex=0 ORDER BY last_login DESC limit 30; 0.02s
SELECT * FROM user WHERE area='美国纽约(Prudential)' AND sex=0 ORDER BY last_login DESC limit 30; 0.04s
SELECT * FROM user WHERE area='印度尼西亚' AND sex=0 ORDER BY last_login DESC limit 30; 0.06s

综上所述：1.建立索引不一定能够加快查询效率如sex这种给重复次数特别多的列增加索引如sex这种会降低查询效率，具体的原因有待查找
2.给重复次数比较少的列增加u讴吟还是能够大幅度提高效率
3.给where和orderby之后的字段添加索引才会加快查询效率
4.为每一个列单独建立索引，不能将索引的效率最大化，应该使用索引合并策略，即根据查询条件，建立联合索引
5.联合索引的顺序问题：将选择性高的索引放到前面
6.根据资料建立索引意味着索引按照最左列进行排序，然后事第二列，以此类推。如（last_login ,area）就会按照last_login进行排序，然后才是area
7.根据这次的这个查询条件来说最好的索引是：ALTER TABLE userADD INDEX index_last_login_area (last_login,area)。