mysql中groupby多个字段的的情况分析(来源网络)

[mysql] 2022-01-23 圈点850

摘要:mysql中groupby多个字段的的情况分析(来源网络)

假设有一个业务场景,需要查询用户登录记录信息,其中表结构如下:


[sql] 

CREATE TABLE `tb` (  

  `id` int(11) NOT NULL AUTO_INCREMENT,  

  `uid` int(11) NOT NULL,  

  `ip` varchar(16) NOT NULL,  

  `login_time` datetime,  

  PRIMARY KEY (`id`),  

  KEY (`uid`)  

);  

再来点测试数据:

[sql]

INSERT INTO tb SELECT null, 1001, '192.168.1.1', '2017-01-21 16:30:47';  

INSERT INTO tb SELECT null, 1003, '192.168.1.153', '2017-01-21 19:30:51';  

INSERT INTO tb SELECT null, 1001, '192.168.1.61', '2017-01-21 16:50:41';  

INSERT INTO tb SELECT null, 1002, '192.168.1.31', '2017-01-21 18:30:21';  

INSERT INTO tb SELECT null, 1002, '192.168.1.66', '2017-01-21 19:12:32';  

INSERT INTO tb SELECT null, 1001, '192.168.1.81', '2017-01-21 19:53:09';  

INSERT INTO tb SELECT null, 1001, '192.168.1.231', '2017-01-21 19:55:34';  

表数据情况:

[plain]  

+----+------+---------------+---------------------+  

| id | uid  | ip            | login_time          |  

+----+------+---------------+---------------------+  

| 1  | 1001 | 192.168.1.1   | 2017-01-21 16:30:47 |  

| 2  | 1003 | 192.168.1.153 | 2017-01-21 19:30:51 |  

| 3  | 1001 | 192.168.1.61  | 2017-01-21 16:50:41 |  

| 4  | 1002 | 192.168.1.31  | 2017-01-21 18:30:21 |  

| 5  | 1002 | 192.168.1.66  | 2017-01-21 19:12:32 |  

| 6  | 1001 | 192.168.1.81  | 2017-01-21 19:53:09 |  

| 7  | 1001 | 192.168.1.231 | 2017-01-21 19:55:34 |  

+----+------+---------------+---------------------+  

如果只需要针对用户查出其最后登录的时间,可以简单写出:

[html]  

SELECT uid, max(login_time)  

FROM tb  

GROUP BY uid;  

[plain]  

+------+---------------------+  

| uid  | max(login_time)       |  

+------+---------------------+  

| 1001 | 2017-01-21 19:55:34 |  

| 1002 | 2017-01-21 19:12:32 |  

| 1003 | 2017-01-21 19:30:51 |  

+------+---------------------+  

若还需要查询用户最后登录时的其他信息,就不能用这种sql写了:

[sql]  

-- 错误写法  

SELECT uid, ip, max(login_time)  

FROM tb  

GROUP BY uid;  

-- 错误写法  

这样的语句是非SQL标准的,虽然能够在MySQL数据库中执行成功,但返回的却是未知的

(如果sql_mode开启了only_full_group_by,则不会执行成功。)

可能ip字段会取uid分组前的第一个row的值,显然不是所需信息

写法1

写一个子查询:

[sql]  

SELECT a.uid, a.ip, a.login_time  

FROM tb a  

WHERE a.login_time in (  

SELECT max(login_time)  

FROM tb  

GROUP BY uid);  

写法2

再或者换一个写法:

[sql]  

SELECT a.uid, a.ip, a.login_time  

FROM tb a  

WHERE a.login_time = (  

SELECT max(login_time)  

FROM tb  

WHERE a.uid = uid);  

顺便测了一下

在5.6以前的版本中,写法②这条sql在大数据量的情况下,执行计划不理想,目测性能不佳。

在5.6及以后的版本中,写法②这条sql会快很多,执行计划也有了改变

5.5.50:

[plain]  

+----+--------------------+-------+------+---------------+------+---------+------+------+-------------+  

| id | select_type        | table | type | possible_keys | key  | key_len | ref  | rows | Extra       |  

+----+--------------------+-------+------+---------------+------+---------+------+------+-------------+  

| 1  | PRIMARY            | a     | ALL  | NULL             | NULL  | NULL      | NULL | 7    | Using where |  

| 2  | DEPENDENT SUBQUERY | tb    | ALL  | uid           | NULL  | NULL      | NULL | 7    | Using where |  

+----+--------------------+-------+------+---------------+------+---------+------+------+-------------+  

5.6.30:

[plain]  

+----+--------------------+-------+------+---------------+------+---------+------------+------+-------------+  

| id | select_type        | table  | type | possible_keys | key  | key_len | ref       | rows  | Extra      |  

+----+--------------------+-------+------+---------------+------+---------+------------+------+-------------+  

| 1  | PRIMARY            | a     | ALL  | NULL              | NULL | NULL      | NULL        | 7    | Using where |  

| 2  | DEPENDENT SUBQUERY | tb    | ref  | uid           | uid  | 4       | test.a.uid | 1    | NULL           |  

+----+--------------------+-------+------+---------------+------+---------+------------+------+-------------+  

写法3

直接改成join性能会更加好:

[sql]  

SELECT a.uid, a.ip, a.login_time  

FROM (SELECT uid, max(login_time) login_time  

FROM tb  

GROUP BY uid  

) b JOIN tb a ON a.uid = b.uid AND a.login_time = b.login_time;  

当然,结果都相同:

[plain]  

+------+---------------+---------------------+  

| uid  | ip            | login_time          |  

+------+---------------+---------------------+  

| 1003 | 192.168.1.153 | 2017-01-21 19:30:51 |  

| 1002 | 192.168.1.66  | 2017-01-21 19:12:32 |  

| 1001 | 192.168.1.231 | 2017-01-21 19:55:34 |  

+------+---------------+---------------------+  

注:如果要分组取最小值直接改对应函数和符号就行了。

groupby  多个字段  

感谢反馈,已提交成功,审核后即会显示