MySQL join 连表查询索引问题
首先先创建两个临时表,并加一条基础数据进去
create table user ( id int auto_increment comment '自增主键' primary key, name varchar(30) null comment '用户名称', create_time datetime not null comment '注册时间', last_login_time datetime null comment '最后登录时间' ) comment '测试表'; create table article ( id int auto_increment comment '自增主键' primary key, user_id int not null comment '用户id', name varchar(30) null comment '帖子名称', topic varchar(30) null comment '帖子关键词', content varchar(500) null comment '帖子内容', create_time datetime not null comment '创建时间' ) comment '测试表2';*/insert into article values(1,1, 'name_1', 'topic_1', 'content_1', '2019-01-01 00:00:00'); insert into user values(1,'user_1', '2019-01-01 00:00:00', '2019-03-01 12:00:00');
为了能模拟大查询的情况,给每个表插入一些数据,user要有万级数据量,article要有百万级数据量,下面的sql每执行一次,数据量翻倍,谨慎执行!
set @i=1; set @time=1; insert into user(name, create_time, last_login_time) select concat('user_',@i:=@i+1), date_add(create_time,interval +@time*cast(rand()*100 as signed) SECOND), null from user; select count(1) from user; set @i=1; set @time=1; insert into article(user_id, name, topic, content, create_time) select round(rand()*(select max(id) from user)), concat('name_',@i:=@i+1), concat('topic_',@i:=@i+1), concat('content_',@i:=@i+1), date_add(create_time,interval +@time*cast(rand()*100 as signed) SECOND) from article; select count(1) from article;
看下查询的SQL语句
select sql_no_cache * from user left join article on(user.id = article.user_id) where user.name like 'user_4%';
没有使用缓存,user表的id是主键,article表除主键外没有任何索引,这种情况下,百万级数据查询情况如下
sql> select sql_no_cache * from user left join article on(user.id = article.user_id) where user.name like 'user_4%' [2020-05-17 13:24:45] 500 rows retrieved starting from 1 in 4 s 681 ms (execution: 1 s 312 ms, fetching: 3 s 369 ms)
给article表加个索引
CREATE INDEX user_id ON article (user_id);
再执行一下看下效果
sql> select sql_no_cache * from user left join article on(user.id = article.user_id) where user.name like 'user_4%' [2020-05-17 13:27:22] 500 rows retrieved starting from 1 in 142 ms (execution: 112 ms, fetching: 30 ms)
可以看出,加了索引后,时间缩短了97%
结论:
A表与B表使用Join联合查询的时候,针对on里面的字段加与不加索引的效率(假设on的条件是A.id=B.aid):
1、两个字段都不加索引,效率极低
2、A表的字段加了索引,B表的字段没有加索引,效果同上
3、A表B表的字段都加了索引,效果很明显
4、A表不加索引,B表加了索引,效果同上
参考: