MySQL 8.0开始Group by不再排序

如题所示，mysql 8.0 开始 group by 默认是没有排序的
那mysql 8.0 之前和 8.0 就有可能结果出现不同需要警惕
查看版本信息

1234567891011121314151617
运行如下SQL
表信息

root@mysql3306.sock>[employees]>select * from t_group ;
+--------+---------+------------+------------+
| emp_no | dept_no | from_date | to_date |
+--------+---------+------------+------------+
| 22744 | d006 | 1986-12-01 | 9999-01-01 |
| 24007 | d005 | 1986-12-01 | 9999-01-01 |
| 30970 | d005 | 1986-12-01 | 2017-03-29 |
| 31112 | d002 | 1986-12-01 | 1993-12-10 |
| 40983 | d005 | 1986-12-01 | 9999-01-01 |
| 49554 | d008 | 1986-12-01 | 1992-05-27 |
| 48317 | d008 | 1986-12-01 | 1989-01-11 |
| 49667 | d007 | 1986-12-01 | 9999-01-01 |
| 50449 | d005 | 1986-12-01 | 9999-01-01 |
| 10004 | d004 | 1986-12-01 | 9999-01-01 |
+--------+---------+------------+------------+
10 rows in set (0.00 sec)

查看执行计划发现extra 部分含有 usisng fiesort

运行结果是按照 dept_no 排序输出

root@mysql3306.sock>[employees]>select dept_no ,count(*) from t_group group by dept_no ;
+---------+----------+
| dept_no | count(*) |
+---------+----------+
| d002 | 1 |
| d004 | 1 |
| d005 | 4 |
| d006 | 1 |
| d007 | 1 |
| d008 | 2 |
+---------+----------+
6 rows in set (0.00 sec)

root@mysql3308.sock>[test]>select * from t_group ;
+--------+---------+------------+------------+
| emp_no | dept_no | from_date | to_date |
+--------+---------+------------+------------+
| 22744 | d006 | 1986-12-01 | 9999-01-01 |
| 24007 | d005 | 1986-12-01 | 9999-01-01 |
| 30970 | d005 | 1986-12-01 | 2017-03-29 |
| 31112 | d002 | 1986-12-01 | 1993-12-10 |
| 40983 | d005 | 1986-12-01 | 9999-01-01 |
| 49554 | d008 | 1986-12-01 | 1992-05-27 |
| 48317 | d008 | 1986-12-01 | 1989-01-11 |
| 49667 | d007 | 1986-12-01 | 9999-01-01 |
| 50449 | d005 | 1986-12-01 | 9999-01-01 |
| 10004 | d004 | 1986-12-01 | 9999-01-01 |
+--------+---------+------------+------------+

1234567891011121314151617181920212223242526272829303132
运行相同的sql
执行计划可以看出 extra 部分已经没有 using filesort

执行结果中发现不是按照 dept_no 排序输出的

root@mysql3308.sock>[test]>select dept_no ,count(*) from t_group group by dept_no ;
+---------+----------+
| dept_no | count(*) |
+---------+----------+
| d006 | 1 |
| d005 | 4 |
| d002 | 1 |
| d008 | 2 |
| d007 | 1 |
| d004 | 1 |
+---------+----------+
6 rows in set (0.00 sec)

12345678910111213141516171819202122232425
如果这种情况下进行分页
5.7 版本
root@mysql3306.sock>[employees]>select dept_no ,count(*) from t_group group by dept_no limit 1;
+---------+----------+
| dept_no | count(*) |
+---------+----------+
| d002 | 1 |
+---------+----------+
1 row in set (0.00 sec)
1234567
8.0版本
root@mysql3308.sock>[test]>select dept_no ,count(*) from t_group group by dept_no limit 1;
+---------+----------+
| dept_no | count(*) |
+---------+----------+
| d006 | 1 |
+---------+----------+
1 row in set (0.00 sec)
1234567
可以发现两个结果是不一样的，原因就是8.0之前是group by 包含using filesort
而 8.0开始是没有的
为了避免这种问题，8.0开始除了索引替换排序的优化思路之外的相关排序我们必须使用order by来进行
5.7
root@mysql3306.sock>[employees]>select dept_no ,count(*) from t_group group by dept_no order by dept_no limit 1;
+---------+----------+
| dept_no | count(*) |
+---------+----------+
| d002 | 1 |
+---------+----------+
1 row in set (0.00 sec)

8.0
root@mysql3308.sock>[test]>select dept_no ,count(*) from t_group group by dept_no order by dept_no limit 1;
+---------+----------+
| dept_no | count(*) |
+---------+----------+
| d002 | 1 |
+---------+----------+
1 row in set (0.00 sec)
1234567891011121314151617
我是知数堂SQL 优化班老师~ ^^
---------------------
作者：骑龟的兔子2018
来源：CSDN
原文：https://blog.csdn.net/qidan3500/article/details/84563296
版权声明：本文为博主原创文章，转载请附上博文链接！