SpringBoot & Sharding-JDBC MasterSlave Example

概述

读写分离，简单来说，就是将DML交给主数据库去执行，将更新结果同步至各个从数据库保持主从数据一致，DQL分发给从数据库去查询，从数据库只提供读取查询操作。读写分离特别适用于读多写少的场景下，通过分散读写到不同的数据库实例上来提高性能，缓解单机数据库的压力:

Name	Remark
DQL	数据查询语言，比如select查询语句
DML	数据操纵语言，比如insert、delete、update更新语句
DDL	数据定义语言，比如create/drop/alter等语句
DCL	数据控制语言，比如grant/rollback/commit等语句

Sharding-JDBC是一个开源的分布式数据库中间件解决方案。它在Java的JDBC层以对业务应用零侵入的方式额外提供数据分片，读写分离，柔性事务和分布式治理能力。并在其基础上提供封装了MySQL协议的服务端版本，用于完成对异构语言的支持。

基于JDBC的客户端版本定位为轻量级Java框架，使用客户端直连数据库，以jar包形式提供服务，无需额外部署和依赖，可理解为增强版的JDBC驱动，完全兼容JDBC和各种ORM框架。

封装了MySQL协议的服务端版本定位为透明化的MySQL代理端，可以使用任何兼容MySQL协议的访问客户端(如：MySQL Command Client, MySQL Workbench等)操作数据，对DBA更加友好。
以上内容摘抄至Sharding-JDBC官网 (http://shardingjdbc.io/docume...

本文主要探讨在SpringBoot环境下如何使用Sharding-JDBC提供的读写分离解决方案;

环境

SpringBoot: 1.5.7.RELEASE,
MybatisPlus: 2.1.4,
Sharding-JDBC: 2.0.0.M2

POM.xml

Sharding-JDBC现已提供相关的Starter, 集成起来非常简单;

下面是完整的pom文件(springboot & mysql & mybatis-plus & sharding-jdbc)

    4.0.0

    com.example
    sharding-jdbc-example-with-spring-boot
    0.0.1-SNAPSHOT
    jar

    sharding-jdbc-example-with-spring-boot
    Demo project for Spring Boot

    
        org.springframework.boot
        spring-boot-starter-parent
        1.5.7.RELEASE
    

    
        UTF-8
        1.7
        1.7
        1.0.5
        2.1.4
        2.7.2
        1.2.39
        1.4
        5.1.30
    

    
        
            org.springframework.boot
            spring-boot-starter-web
        
        
            org.springframework.boot
            spring-boot-starter-jetty
        
        
            com.h2database
            h2
        
        
            com.zaxxer
            HikariCP
            ${HikariCP.version}
        

        
        
            com.baomidou
            mybatisplus-spring-boot-starter
            ${mybatisplus-spring-boot-starter.version}
        
        
            com.baomidou
            mybatis-plus
            ${mybatisplus.version}
        
        

        
        
            org.springframework.boot
            spring-boot-starter-test
        
        
            com.alibaba
            fastjson
            ${fastjson.version}
        
        
        
            io.shardingjdbc
            sharding-jdbc-spring-boot-starter
            2.0.0.M2
        
        
            commons-dbcp
            commons-dbcp
            ${commons-dbcp.version}
        
        
            mysql
            mysql-connector-java
            ${mysql-connector-java.version}
        

        
            org.projectlombok
            lombok
            true
        
        
        
            cn.hutool
            hutool-all
            4.0.12
        
    

    
        
            
                org.springframework.boot
                spring-boot-maven-plugin
            
            
                org.apache.maven.plugins
                maven-compiler-plugin
                
                    8
                    8
                
            
        
    

多数据源配置 (application.yml)

server:
  port: 10086

sharding:
  jdbc:
      datasource:
        names: ds_master_0,ds_slave_0_1,ds_slave_0_2
        ds_master_0:
          type: org.apache.commons.dbcp.BasicDataSource
          driverClassName: com.mysql.jdbc.Driver
          url: jdbc:mysql://127.0.01:3306/ds_master?useUnicode=true&characterEncoding=UTF-8&useSSL=false
          username: root
          password: root
        ds_slave_0_1:
          type: org.apache.commons.dbcp.BasicDataSource
          driverClassName: com.mysql.jdbc.Driver
          url: jdbc:mysql://127.0.01:3306/ds_slave_0?useUnicode=true&characterEncoding=UTF-8&useSSL=false
          username: root
          password: root
        ds_slave_0_2:
          type: org.apache.commons.dbcp.BasicDataSource
          driverClassName: com.mysql.jdbc.Driver
          url: jdbc:mysql://127.0.01:3306/ds_slave_1?useUnicode=true&characterEncoding=UTF-8&useSSL=false
          username: root
          password: root
      config:
        # 主从策略
        masterslave:
          load-balance-algorithm-type: round_robin # 负载策略
          name: ds_m_1_s_2
          master-data-source-name: ds_master_0
          slave-data-source-names: ds_slave_0_1,ds_slave_0_2
        sharding:
          props:
            sql:
              show: true

#mybatis
mybatis-plus:
  datasource: dataSource
  mapper-locations: classpath:/mapper/*Mapper.xml
  #实体扫描，多个package用逗号或者分号分隔
  typeAliasesPackage: com.example.shardingjdbcexamplewithspringboot.entity
  global-config:
    #主键类型  0:"数据库ID自增", 1:"用户输入ID",2:"全局唯一ID (数字类型唯一ID)", 3:"全局唯一ID UUID";
    id-type: 0
    #字段策略 0:"忽略判断",1:"非 NULL 判断"),2:"非空判断"
    field-strategy: 2
    #驼峰下划线转换
    db-column-underline: true
    #刷新mapper 调试神器
    refresh-mapper: true
    #逻辑删除配置
    logic-delete-value: 0
    logic-not-delete-value: 1
  configuration:
    map-underscore-to-camel-case: true
    cache-enabled: false

实际上到这一步就完成了最简单的配置, 为了测试效果, 生成相关的数据库和实体吧:

init data.sql;

建立3个数据库, 分别为主, 从1, 从2, 为了区分数据来源, 在from中指定了节点名称;

CREATE SCHEMA IF NOT EXISTS `ds_master`;

DROP TABLE IF EXISTS `ds_master`.`tb_employee`;
CREATE TABLE `ds_master`.`tb_employee` (
`id`  int NOT NULL AUTO_INCREMENT ,
`name`  varchar(255) NULL ,
`from`  varchar(255) NULL ,
PRIMARY KEY (`id`)
);

INSERT INTO `ds_master`.`tb_employee` VALUES(1,'name1', 'ds_master');
INSERT INTO `ds_master`.`tb_employee` VALUES(2,'name2', 'ds_master');
INSERT INTO `ds_master`.`tb_employee` VALUES(3,'name3', 'ds_master');

####
CREATE SCHEMA IF NOT EXISTS `ds_slave_0`;

DROP TABLE IF EXISTS `ds_slave_0`.`tb_employee`;
CREATE TABLE `ds_slave_0`.`tb_employee` (
`id`  int NOT NULL AUTO_INCREMENT ,
`name`  varchar(255) NULL ,
`from`  varchar(255) NULL ,
PRIMARY KEY (`id`)
);

INSERT INTO `ds_slave_0`.`tb_employee` VALUES(1,'name1', 'ds_slave_0');
INSERT INTO `ds_slave_0`.`tb_employee` VALUES(2,'name2', 'ds_slave_0');
INSERT INTO `ds_slave_0`.`tb_employee` VALUES(3,'name3', 'ds_slave_0');


####
CREATE SCHEMA IF NOT EXISTS `ds_slave_1`;

DROP TABLE IF EXISTS `ds_slave_1`.`tb_employee`;
CREATE TABLE `ds_slave_1`.`tb_employee` (
`id`  int NOT NULL AUTO_INCREMENT ,
`name`  varchar(255) NULL ,
`from`  varchar(255) NULL ,
PRIMARY KEY (`id`)
);

INSERT INTO `ds_slave_1`.`tb_employee` VALUES(1,'name1', 'ds_slave_1');
INSERT INTO `ds_slave_1`.`tb_employee` VALUES(2,'name2', 'ds_slave_1');
INSERT INTO `ds_slave_1`.`tb_employee` VALUES(3,'name3', 'ds_slave_1');

Entity / Mapper / Service:

@Data
@Builder
@ToString
@NoArgsConstructor
@AllArgsConstructor
@TableName("tb_employee")
public class EmployeeEntity {

    @TableId(type = IdType.AUTO)
    private Integer id;

    @TableField
    private String name;

    @TableField
    private String from;
}


public interface EmployeeMapper extends BaseMapper {
}

单元测试

@RunWith(SpringRunner.class)
@SpringBootTest(classes = ShardingJdbcExampleWithSpringBootApplication.class)
public class ShardingJdbcExampleWithSpringBootApplicationTests {

    @Resource
    EmployeeMapper employeeMapper;


    @Test
    public void testMasterSlave() {
        // search slave db;
        Console.log("search slave db:");
        for (int i = 0; i < 4; i++) {
            ((Runnable) () -> {
                Console.log(employeeMapper.selectById(1));
            }).run();
        }
        Console.log("==========================================\n");

        EmployeeEntity employeeEntity = EmployeeEntity.builder().name("test").from("write master db").build();
        Integer insert = employeeMapper.insert(employeeEntity);
        Console.log("write master db: {}", insert > 0); // true
        Console.log("==========================================\n");

        EmployeeEntity ret = employeeMapper.selectOne(employeeEntity);
        Console.log("search by \"write master db\": {}", ret); // null
        Console.log("==========================================\n");

           //  强制路由,访问masterdb数据
        HintManager hintManager = HintManager.getInstance();
        hintManager.setMasterRouteOnly();
        ret = employeeMapper.selectOne(employeeEntity); //(id=9, name=test, from=write master db)
        Console.log("[HintManager]search by \"write master db\": {}", ret);
        hintManager.close();
    }
}

最终结果

search slave db:
EmployeeEntity(id=1, name=name1, from=ds_slave_0)
EmployeeEntity(id=1, name=name1, from=ds_slave_0)
EmployeeEntity(id=1, name=name1, from=ds_slave_1)
EmployeeEntity(id=1, name=name1, from=ds_slave_0)
==========================================

write master db: true
==========================================

search by "write master db": null
==========================================

[HintManager]search by "write master db": EmployeeEntity(id=11, name=test, from=write master db)

如测试效果一般, Sharding-JDBC可以帮你轻松的实现读写分离, 但是数据同步仍然是需要考虑的问题;