记踩到 MySQL in 子查询的“坑”

MySQL是项目中常用的数据库，其中in查询也是很常用。最近项目调试过程中，遇到一个出乎意料的select查询，竟然用了33秒！

1. userinfo 表

2. article 表

select*fromuserinfowhereidin(selectauthor_idfromartilcewheretype=1);

大家第一眼看到上面的SQL时，可能都会觉得这是一个很简单的子查询。先把author_id查出来，再用in查询一下。

如果有相关索引会非常快的，拆解来讲就是以下这样的：

1.selectauthor_idfromartilcewheretype=1;　　2.select*fromuserinfowhereidin(1,2,3);

但是事实是这样的：

mysql> select count(*) from userinfo;

mysql> select count(*) from article;

mysql> select id,username from userinfo where id in (select author_id from article where type = 1);

33 秒！为什么会这么慢呢？

官方文档解释：in 子句在查询的时候有时会被转换为 exists 的方式来执行，变成逐条记录进行遍历（版本 5.5 中存在，5.6 中已做优化）。

参考：

https://dev.mysql.com/doc/refman/5.5/en/subquery-optimization.html

1. 使用临时表

select id,username from userinfo
where id in (select author_id from
(select author_id from article where type = 1) as tb);

2. 使用 join

select a.id,a.username from userinfo a, article b
where a.id = b.author_id and b.type = 1;

版本 5.6 已针对子查询做了优化，方式跟【四】中的临时表方式一样，参考官方文档：

If materialization is not used, the optimizer sometimes rewrites a noncorrelated subquery as a correlated subquery.

For example, the following IN subquery is noncorrelated  ( where_condition involves only columns from t2 and not t1 ):

select * from t1

where t1.a in (select t2.b from t2 where where_condition);

The optimizer might rewrite this as an EXISTS correlated subquery:

select * from t1

where exists (select t2.b from t2 where where_condition and t1.a=t2.b);

Subquery materialization using a temporary table avoids such rewrites and makes it possible to execute the subquery only once rather than once per row of the outer query.

https://dev.mysql.com/doc/refman/5.6/en/subquery-materialization.html

文章来自微信公众号：HULK一线技术杂谈