最近有个业务需要能够存储Json并做一些简单的业务逻辑处理。业务找到我说json的数据分析很难用mysql 5.6,这样的纯粹行存来处理难度很大，问我有没啥办法。

我第一想到的是mongodb，第二想到的就是mysql 5.7 。 然后一查，哎呀，已经GA了。众所周知的，mongodb的引擎层的稳定性一直是短板，而innodb经过10年的验证，已经是非常稳定的东西了。

所以就在想，是不是可以试试mysql 5.7 ， 简单尝试，发现mysql 5.7+ DRDS ，完全不带mongoDB玩啊。。。

看看我的简单功能测试：

create table json_test ( uid int auto_increment,datajson,primary key(uid))engine=innodb;

建库

mysql> insert into json_test values (NULL,'{"name":"name1","mobile":"15044447279","amount":400}');

Query OK, 1 row affected (0.01 sec)

mysql> insert into json_test values (NULL,'{"name":"name1","mobile":"15044447279","amount":300}');

Query OK, 1 row affected (0.01 sec)

mysql> insert into json_test values (NULL,'{"name":"name2","mobile":"15044447278","amount":300}');

Query OK, 1 row affected (0.01 sec)

mysql> insert into json_test values (NULL,'{"name":"name3","mobile":"15044447277","amount":300}');

Query OK, 1 row affected (0.01 sec)

>

插入四条语句

mysql> select data from json_test;

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

| data                                             |

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

| {"name": "name1", "amount": 400, "mobile": "15044447279"}|

| {"name": "name1", "amount": 300, "mobile": "15044447279"}|

| {"name": "name2", "amount": 300, "mobile": "15044447278"}|

| {"name": "name3", "amount": 300, "mobile": "15044447277"}|

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

>

查询这四条json语句

mysql> select data->"$.name" as name,sum(data->"$.amount") from json_test group by name;

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

| name    |sum(data->"$.amount") |

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

| "name1" |                700 |

| "name2" |                300 |

| "name3" |                300 |

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

>

做个group by sum 常见的统计操作

insert into json_test values (NULL,'{"mobile":"15044447277","amount":300}');

>

插入一个不带name的数据，看看索引对空数据的兼容性情况。

mysql> ALTER TABLE json_test  ADDuser_name varchar(128) GENERATED ALWAYS AS(json_extract(data,'$.name')) VIRTUAL;

Query OK, 0 rows affected (0.05 sec)

Records: 0  Duplicates: 0 Warnings: 0

mysql> alter table json_test add index idx_username(user_name);

Query OK, 0 rows affected (0.04 sec)

Records: 0  Duplicates: 0 Warnings: 0

>

创建虚拟列并建立索引

mysql> select user_name,sum(data->"$.amount") fromjson_test where user_name = '"name1"';

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

| user_name | sum(data->"$.amount") |

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

| "name1"   |                700 |

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

>

查看符合某个user_name的数据的sum。

mysql> explain select user_name,sum(data->"$.amount")from json_test where user_name = '"name1"';

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

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

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

|  1 | SIMPLE     | json_test | NULL      | ref | idx_username  | idx_username |131     | const |   2 |   100.00| NULL  |

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

>

确认带索引数据走到了索引上

===========

然后，这东西竟然还支持事务。。这个就牛逼大了。。

mysql> start transaction;

Query OK, 0 rows affected (0.00 sec)

>

开启事务

mysql> select * from json_test;

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

| uid | data                                                    | user_name |

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

|   1 | {"name": "name1", "amount": 400,"mobile": "15044447279"}        | "name1"  |

|   2 | {"name": "name1", "amount": 300,"mobile": "15044447279"}        | "name1"  |

|   3 | {"name": "name2", "amount": 300,"mobile": "15044447278"}        | "name2"  |

|   4 | {"name": "name3", "amount": 300,"mobile": "15044447277"}        | "name3"  |

|   5 | {"amount": 300, "mobile":"15044447277"}                      | NULL     |

|   6 | {"amount": "300","name”:”name2”,”mobile": "15044447278"}        | NULL     |

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

>

查看原表。

mysql> insert into json_test (uid,data) values (NULL,'{"name":"name1","mobile":"15044447279","amount":300}');

Query OK, 1 row affected (0.00 sec)

>

插入新数据

mysql> select * from json_test;                                                                                                                                               +-----+-------------------------------------------------------------------+-----------+

| uid | data                                                    | user_name |

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

|   1 | {"name": "name1", "amount": 400,"mobile": "15044447279"}        | "name1"  |

|   2 | {"name": "name1", "amount": 300,"mobile": "15044447279"}        | "name1"  |

|   3 | {"name": "name2", "amount": 300,"mobile": "15044447278"}        | "name2"  |

|   4 | {"name": "name3", "amount": 300,"mobile": "15044447277"}        | "name3"  |

|   5 | {"amount": 300, "mobile":"15044447277"}                      | NULL     |

|   6 | {"amount": "300","name”:”name2”,”mobile": "15044447278"}        | NULL     |

|   7 | {"name": "name1", "amount": 300,"mobile": "15044447279"}        | "name1"  |

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

7 rows in set (0.00 sec)

>

确认新数据

mysql> rollback;

回滚数据

mysql> select * from json_test;

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

| uid | data                                                    | user_name |

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

|   1 | {"name": "name1", "amount": 400,"mobile": "15044447279"}        | "name1"  |

|   2 | {"name": "name1", "amount": 300,"mobile": "15044447279"}        | "name1"  |

|   3 | {"name": "name2", "amount": 300,"mobile": "15044447278"}        | "name2"  |

|   4 | {"name": "name3", "amount": 300,"mobile": "15044447277"}        | "name3"  |

|   5 | {"amount": 300, "mobile":"15044447277"}                      | NULL     |

|   6 | {"amount": "300","name”:”name2”,”mobile": "15044447278"}        | NULL     |

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

6 rows in set (0.00 sec)

>

对mysql 5.7 好感度*2啊。。卧槽。

猜测一下虚拟列的做法：

ALTER TABLE json_test  ADD user_namevarchar(128) GENERATED ALWAYS AS (json_extract(data,'$.name'))VIRTUAL;

>

应该写入的时候做一个trigger 每个json都运算json_extract(data,'$.name')，然后写到一个新的不可修改的列里。

这个就可以让原来的行存和文档有一个非常完美的结合，当业务变化大的时候，放到json里面，而当变化稳定下来，就迁移到行存里。

完美。 推荐！

微博_ID:淘宝沈询_WhisperXD

http://weibo.com/p/230418693f08470102wi35