MySQL使用in带子查询的时候,子查询不要使用union或union all
特别是当外部表比较大的时候,千万不要使用in和union搭配,因为子查询中一旦使用union,执行计划会出现dependent subquery这种情况,
在生产上我们有使用类似的情况,导致SQL执行效率很差,下面举例说明,为了生产安全隐私,以下举例用测试表演示,原理相通。
举例
(1) 使用in和union搭配的时候,s表作为外部表,全表扫描,有260w行,执行20多秒。
mysql> select s.* from salaries s where s.emp_no in (select emp_no from employees e where e.first_name='Georgi' union all select emp_no from employees e where e.hire_date='1992-12-18');
2718 rows in set (21.14 sec)
mysql> desc select s.* from salaries s where s.emp_no in (select emp_no from employees e where e.first_name='Georgi' union all select emp_no from employees e where e.hire_date='1992-12-18');
+----+--------------------+-------+------------+--------+---------------+---------+---------+------+---------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+--------------------+-------+------------+--------+---------------+---------+---------+------+---------+----------+-------------+
| 1 | PRIMARY | s | NULL | ALL | NULL | NULL | NULL | NULL | 2612229 | 100.00 | Using where |
| 2 | DEPENDENT SUBQUERY | e | NULL | eq_ref | PRIMARY | PRIMARY | 4 | func | 1 | 10.00 | Using where |
| 3 | DEPENDENT UNION | e | NULL | eq_ref | PRIMARY | PRIMARY | 4 | func | 1 | 10.00 | Using where |
+----+--------------------+-------+------------+--------+---------------+---------+---------+------+---------+----------+-------------+
3 rows in set, 1 warning (0.00 sec)
(2)可以使用join来转化,再来看执行计划e表变成外表,s表使用PK检索,执行只要了0.32秒,效率大大提高。
mysql> select s.* from salaries s join (select emp_no from employees e where e.first_name='Georgi' union all select emp_no from employees e where e.hire_date='1992-12-18')e on s.emp_no=e.emp_no;
2718 rows in set (0.32 sec)
mysql> desc select s.* from salaries s join (select emp_no from employees e where e.first_name='Georgi' union all select emp_no from employees e where e.hire_date='1992-12-18')e on s.emp_no=e.emp_no;
+----+-------------+------------+------------+------+----------------+---------+---------+----------+--------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+------+----------------+---------+---------+----------+--------+----------+-------------+
| 1 | PRIMARY | <derived2> | NULL | ALL | NULL | NULL | NULL | NULL | 59866 | 100.00 | NULL |
| 1 | PRIMARY | s | NULL | ref | PRIMARY,emp_no | PRIMARY | 4 | e.emp_no | 9 | 100.00 | NULL |
| 2 | DERIVED | e | NULL | ALL | NULL | NULL | NULL | NULL | 299335 | 10.00 | Using where |
| 3 | UNION | e | NULL | ALL | NULL | NULL | NULL | NULL | 299335 | 10.00 | Using where |
+----+-------------+------------+------------+------+----------------+---------+---------+----------+--------+----------+-------------+
4 rows in set, 1 warning (0.00 sec)