使用Sharding-JDBC实现数据分片和读写分离

一、Sharding-JDBC简介

Sharding-JDBC是Sharding-Sphere的一个产品，它有三个产品，分别是Sharding-JDBC、Sharding-Proxy和Sharding-Sidecar，这三个产品提供了标准化的数据分片、读写分离、柔性事务和数据治理功能。我们这里用的是Sharding-JDBC，所以想了解后面两个产品的话可以去它们官网查看。

Sharding-JDBC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，兼容性特别强。适用的ORM框架有JPA, Hibernate, Mybatis, Spring JDBC Template或直接使用JDBC；第三方的数据库连接池有DBCP, C3P0, BoneCP, Druid等；支持的数据库有MySQL，Oracle，SQLServer和PostgreSQL；多样化的配置文件Java，yaml，Spring Boot ，Spring命名空间。其实这里说的都是废话，大家可以不看，下面我们动手开始正式配置。

二、具体的实现方式

1、maven引用

我这里用的配置方式是Spring命名空间配置，所以只需要引用sharding-jdbc-spring-namespace就可以了，还有要注意的是我用的不是当当网的sharding，注意groupId是io.shardingsphere。如果用的是其它配置方式可以去http://maven.aliyun.com/nexus/#nexus-search;quick~io.shardingsphere网站查找相应maven引用

<dependency><groupId>io.shardingsphere</groupId><artifactId>sharding-jdbc-spring-namespace</artifactId><version>3.0.0.M1</version>
</dependency>

2、数据库准备

我这里用的是mysql数据库，根据我们项目的具体需求，我准备了三个主库和对应的从库。模拟的主库名有master，暂时没有做对应从库，所以对应的从库还是指向master；第二个主库有master_1，对应的从库有master_1_slaver_1，master_1_slave_2；第三个主库有master_2，对应的从库有master_2_slave_1，master_2_slave_2。
数据库中的表也做了分表，下面是对应的mysql截图。

这第一幅图上的主从库都应该在不同的服务器上的，但这里只是为了模拟所以就建在了本地服务器上了。我们读写分离中的写操作只会发生在主库上，从库会自动同步主库上的数据并为读提供数据。数据库的主从复制在上篇博文中做了详细的介绍，大家可以去看看https://blog.csdn.net/shijiemozujiejie/article/details/80766633

这幅图作为我们本来的主库，下面做的分库和分表都是基于这个库中的订单表分的。所以分库中的表只有订单表和订单明细表。

第三幅图截的是第二个主库，里面对订单和订单明细表做了分表操作，具体的分片策略和分片算法下面再做介绍。第三个主表和第二个主表是一样的，所有的从表都和对应的主表是一致的。

2、Spring配置

数据库信息配置文件db.properties配置可以配置两份，分为开发版和测试版，如下：

# master
Master.url=jdbc:mysql://localhost:3306/master?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master.username=root
Master.password=123456
Slave.url=jdbc:mysql://localhost:3306/master?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Slave.username=root
Slave.password=123456# maste_1
Master_1.url=jdbc:mysql://localhost:3306/master_1?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master_1.username=root
Master_1.password=123456
Master_1_Slave_1.url=jdbc:mysql://localhost:3306/master_1_slave_1?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master_1_Slave_1.username=root
Master_1_Slave_1.password=123456
Master_1_Slave_2.url=jdbc:mysql://localhost:3306/master_1_slave_2?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master_1_Slave_2.username=root
Master_1_Slave_2.password=123456# master_2
Master_2.url=jdbc:mysql://localhost:3306/master_2?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master_2.username=root
Master_2.password=123456
Master_2_Slave_1.url=jdbc:mysql://localhost:3306/master_2_slave_1?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master_2_Slave_1.username=root
Master_2_Slave_1.password=123456
Master_2_Slave_2.url=jdbc:mysql://localhost:3306/master_2_slave_2?useUnicode=true&characterEncoding=utf8&autoReconnect=true&rewriteBatchedStatements=true
Master_2_Slave_2.username=root
Master_2_Slave_2.password=123456

Spring对应的配置：
Spring-Sphere官网中的demo里用的都是行表达式的分片策略，但是行表达式的策略不利于数据库和表的横向扩展，所以我这里用的是标准分片策略，精准分片算法和范围分片算法。因为我们项目中暂时用的分片键都是user_id单一键，所以说不存在复合分片策略，也用不到Hint分片策略，行表达式分片策略和不分片策略。

<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"xmlns:context="http://www.springframework.org/schema/context"xmlns:tx="http://www.springframework.org/schema/tx"xmlns:sharding="http://shardingsphere.io/schema/shardingsphere/sharding"xmlns:master-slave="http://shardingsphere.io/schema/shardingsphere/masterslave"xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context.xsdhttp://www.springframework.org/schema/tx http://www.springframework.org/schema/tx/spring-tx.xsdhttp://shardingsphere.io/schema/shardingsphere/sharding http://shardingsphere.io/schema/shardingsphere/sharding/sharding.xsd http://shardingsphere.io/schema/shardingsphere/masterslave  http://shardingsphere.io/schema/shardingsphere/masterslave/master-slave.xsd"><context:component-scan base-package="com.jihao" /><!-- db.properties数据库信息配置 --><bean id="property" class="org.springframework.beans.factory.config.PropertyPlaceholderConfigurer"><property name="location" value="classpath:property/db_dev.properties" /></bean><!-- 主库 --><bean id="master" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master.url}"/><property name="username" value="${Master.username}"/><property name="password" value="${Master.password}"/></bean><!-- 主库的从库 --><bean id="slave" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Slave.url}"/><property name="username" value="${Slave.username}"/><property name="password" value="${Slave.password}"/></bean><!-- 主库的分库1 --><bean id="master_1" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master_1.url}"/><property name="username" value="${Master_1.username}"/><property name="password" value="${Master_1.password}"/></bean><!-- 分库1的从库1 --><bean id="master_1_slave_1" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master_1_Slave_1.url}"/><property name="username" value="${Master_1_Slave_1.username}"/><property name="password" value="${Master_1_Slave_1.password}"/></bean><!-- 分库1的从库2 --><bean id="master_1_slave_2" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master_1_Slave_2.url}"/><property name="username" value="${Master_1_Slave_2.username}"/><property name="password" value="${Master_1_Slave_2.password}"/></bean><!-- 主库的分库2 --><bean id="master_2" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master_2.url}"/><property name="username" value="${Master_2.username}"/><property name="password" value="${Master_2.password}"/></bean><!-- 分库2的从库1 --><bean id="master_2_slave_1" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master_2_Slave_1.url}"/><property name="username" value="${Master_2_Slave_1.username}"/><property name="password" value="${Master_2_Slave_1.password}"/></bean><!-- 分库2的从库2 --><bean id="master_2_slave_2" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close"><property name="driverClassName" value="com.mysql.jdbc.Driver"/><property name="url" value="${Master_2_Slave_2.url}"/><property name="username" value="${Master_2_Slave_2.username}"/><property name="password" value="${Master_2_Slave_2.password}"/></bean><!-- 主从关系配置 --><bean id="randomStrategy" class="io.shardingsphere.core.api.algorithm.masterslave.RandomMasterSlaveLoadBalanceAlgorithm" /><master-slave:data-source id="ms_master" master-data-source-name="master" slave-data-source-names="slave" strategy-ref="randomStrategy" /><master-slave:data-source id="ms_master_1" master-data-source-name="master_1" slave-data-source-names="master_1_slave_1, master_1_slave_2" strategy-ref="randomStrategy" /><master-slave:data-source id="ms_master_2" master-data-source-name="master_2" slave-data-source-names="master_2_slave_1, master_2_slave_2" strategy-ref="randomStrategy" /><!-- 分库策略  精确分片算法 --><bean id="preciseDatabaseStrategy" class="com.jihao.algorithm.PreciseModuleDatabaseShardingAlgorithm" /><!-- 分库策略  范围分片算法 --><bean id="rangeDatabaseStrategy" class="com.jihao.algorithm.RangeModuleDatabaseShardingAlgorithm" /><!-- 分表策略  精确分片算法 --><bean id="preciseTableStrategy" class="com.jihao.algorithm.PreciseModuleTableShardingAlgorithm" /><!-- 分表策略  范围分片算法--><bean id="rangeTableStrategy" class="com.jihao.algorithm.RangeModuleTableShardingAlgorithm" /><sharding:standard-strategy id="databaseStrategy" sharding-column="user_id" precise-algorithm-ref="preciseDatabaseStrategy" range-algorithm-ref="rangeDatabaseStrategy" /><!-- 分表策略 --><sharding:standard-strategy id="tableStrategy" sharding-column="user_id" precise-algorithm-ref="preciseTableStrategy" range-algorithm-ref="rangeTableStrategy" /><!-- 行表达式算法 --><!-- <sharding:inline-strategy id="databaseStrategy" sharding-column="user_id" algorithm-expression="demo_ds_ms_$->{user_id % 2}" /><sharding:inline-strategy id="orderTableStrategy" sharding-column="order_id" algorithm-expression="t_order_$->{order_id % 2}" /><sharding:inline-strategy id="orderItemTableStrategy" sharding-column="order_item_id" algorithm-expression="t_order_item_$->{order_item_id % 2}" /> --><sharding:data-source id="shardingDataSource"><sharding:sharding-rule data-source-names="ms_master,ms_master_1,ms_master_2"><sharding:table-rules><sharding:table-rule logic-table="t_order" actual-data-nodes="ms_master_$->{1..2}.t_order_$->{1..3}" database-strategy-ref="databaseStrategy" table-strategy-ref="tableStrategy" generate-key-column-name="order_id"/><sharding:table-rule logic-table="t_order_item" actual-data-nodes="ms_master_$->{1..2}.t_order_item_$->{1..3}" database-strategy-ref="databaseStrategy" table-strategy-ref="tableStrategy" generate-key-column-name="order_item_id"/></sharding:table-rules></sharding:sharding-rule></sharding:data-source><bean id="transactionManager" class="org.springframework.jdbc.datasource.DataSourceTransactionManager"><property name="dataSource" ref="shardingDataSource" /></bean><tx:annotation-driven /><bean id="sqlSessionFactory" class="org.mybatis.spring.SqlSessionFactoryBean"><!-- 用于在控制台打印sql（不需要的可以注释掉这一行） --><property name="configLocation" value="classpath:log/mybatis-config.xml"></property><property name="dataSource" ref="shardingDataSource"/><property name="mapperLocations" value="classpath*:com/jihao/mapper/*.xml"/></bean><bean class="org.mybatis.spring.mapper.MapperScannerConfigurer"><property name="basePackage" value="com.jihao"/><property name="sqlSessionFactoryBeanName" value="sqlSessionFactory"/></bean>
</beans>

3、精准分片算法和范围分片算法的Java代码

标准分片策略，精准分片算法

package com.jihao.algorithm;import io.shardingsphere.core.api.algorithm.sharding.PreciseShardingValue;
import io.shardingsphere.core.api.algorithm.sharding.standard.PreciseShardingAlgorithm;import java.util.Collection;import com.alibaba.fastjson.JSON;/*** 自定义标准分片策略，使用精确分片算法（=与IN）* @author JiHao**/
public class PreciseModuleDatabaseShardingAlgorithm implements PreciseShardingAlgorithm<Long>{@Overridepublic String doSharding(Collection<String> availableTargetNames,PreciseShardingValue<Long> preciseShardingValue) {System.out.println("collection:" + JSON.toJSONString(availableTargetNames) + ",preciseShardingValue:" + JSON.toJSONString(preciseShardingValue));for (String name : availableTargetNames) {// =与IN中分片键对应的值String value = String.valueOf(preciseShardingValue.getValue());// 分库的后缀int i = 1;// 求分库后缀名的递归算法if (name.endsWith("_" + countDatabaseNum(Long.parseLong(value), i))) {return name;}}throw new UnsupportedOperationException();}/*** 计算该量级的数据在哪个数据库* @return*/private String countDatabaseNum(long columnValue, int i){// ShardingSphereConstants每个库中定义的数据量long left = ShardingSphereConstants.databaseAmount * (i-1);long right = ShardingSphereConstants.databaseAmount * i;if(left < columnValue && columnValue <= right){return String.valueOf(i);}else{i++;return countDatabaseNum(columnValue, i);}}}

标准分片策略，范围分片算法

package com.jihao.algorithm;import io.shardingsphere.core.api.algorithm.sharding.RangeShardingValue;
import io.shardingsphere.core.api.algorithm.sharding.standard.RangeShardingAlgorithm;import java.util.ArrayList;
import java.util.Collection;
import java.util.List;import com.alibaba.fastjson.JSON;
import com.google.common.collect.Range;/*** 自定义标准分库策略，使用范围分片算法（BETWEEN AND）* @author JiHao**/
public class RangeModuleDatabaseShardingAlgorithm implements RangeShardingAlgorithm<Long>{@Overridepublic Collection<String> doSharding(Collection<String> availableTargetNames,RangeShardingValue<Long> rangeShardingValue) {System.out.println("Range collection:" + JSON.toJSONString(availableTargetNames) + ",rangeShardingValue:" + JSON.toJSONString(rangeShardingValue));Collection<String> collect = new ArrayList<>();Range<Long> valueRange = rangeShardingValue.getValueRange();// BETWEEN AND中分片键对应的最小值long lowerEndpoint = Long.parseLong(String.valueOf(valueRange.lowerEndpoint()));// BETWEEN AND中分片键对应的最大值long upperEndpoint = Long.parseLong(String.valueOf(valueRange.upperEndpoint()));// 分表的后缀int i = 1;List<Integer> arrs = new ArrayList<Integer>();// 求分表后缀名的递归算法List<Integer> list = countDatabaseNum(i, lowerEndpoint, upperEndpoint, arrs);for (Integer integer : list) {for (String each : availableTargetNames) {if (each.endsWith("_" + integer)) {collect.add(each);}}}return collect;}/*** 计算该量级的数据在哪个表* @param columnValue* @param i* @param lowerEndpoint 最小区间* @param upperEndpoint 最大区间* @return*/private List<Integer> countDatabaseNum(int i, long lowerEndpoint, long upperEndpoint, List<Integer> arrs){long left = ShardingSphereConstants.databaseAmount * (i-1);long right = ShardingSphereConstants.databaseAmount * i;// 区间最大值小于分库最大值if(left < upperEndpoint && upperEndpoint <= right){arrs.add(i);return arrs;}else{if(left < lowerEndpoint && lowerEndpoint <= right){arrs.add(i);}i++;return countDatabaseNum(i, lowerEndpoint, upperEndpoint, arrs);}}}

分库的策略用的和分库的代码是一样的，不同之处就是分库用的是databaseAmount，分表用的是tableAmount。下面的ShardingSphereConstants的代码。

package com.jihao.algorithm;/*** ShardingSphere中用到的常量* @author JiHao**/
public class ShardingSphereConstants {/*** 订单、优惠券相关的表，按用户数量分库，64w用户数据为一个库* (0,64w]*/public static int databaseAmount = 640000;/*** 一个订单表里存10000的用户订单* (0,1w]*/public static int tableAmount = 10000;}

到这里所有的配置基本上都已经完成了，下面的测试。

4、测试

下面是测试的mybatis的测试文件，都是最基础的就不讲解了。

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN" "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
<mapper namespace="com.jihao.dao.TestShardingMapper"><resultMap id="BaseResultMap" type="com.jihao.entity.Order"><id column="order_id" jdbcType="INTEGER" property="orderId" /><result column="user_id" jdbcType="INTEGER" property="userId" /><result column="status" jdbcType="INTEGER" property="status" /></resultMap><insert id="insert" parameterType="com.jihao.entity.Order" useGeneratedKeys="true" keyProperty="orderId">INSERT INTO t_order (user_id, status)VALUES (#{userId,jdbcType=INTEGER}, #{status,jdbcType=VARCHAR})</insert><insert id="insertItem" useGeneratedKeys="true" keyProperty="orderItemId">INSERT INTO t_order_item (order_id, user_id)VALUES (#{orderId,jdbcType=INTEGER},#{userId,jdbcType=INTEGER})</insert><select id="searchOrder" resultMap="BaseResultMap">SELECT * from t_order</select><select id="queryWithEqual" resultMap="BaseResultMap">SELECT * FROM t_order WHERE user_id=51</select><select id="queryWithIn" resultMap="BaseResultMap">SELECT * FROM t_order WHERE user_id IN (50, 51)</select><select id="queryWithBetween" resultMap="BaseResultMap">SELECT * FROM t_order WHERE user_id between 10000 and 30000</select><select id="queryUser" resultType="Map">SELECT * FROM t_user</select></mapper>

下面对应的mapper的Java代码

package com.jihao.dao;import java.util.List;
import java.util.Map;import org.apache.ibatis.annotations.Mapper;import com.jihao.entity.Order;
import com.jihao.entity.OrderItem;@Mapper
public interface TestShardingMapper {int insert(Order record);int insertItem(OrderItem record);List<Order> searchOrder();List<Order> queryWithEqual();List<Order> queryWithIn();List<Order> queryWithBetween();List<Map<String, Object>> queryUser();}

下面是对应的订单entity代码

package com.jihao.entity;/*** 订单* @author JiHao*/
public class Order {private Long orderId;private Integer userId;private String status;public Long getOrderId() {return orderId;}public void setOrderId(Long orderId) {this.orderId = orderId;}public Integer getUserId() {return userId;}public void setUserId(Integer userId) {this.userId = userId;}public String getStatus() {return status;}public void setStatus(String status) {this.status = status;}}

下面是对应的订单明细entity代码

package com.jihao.entity;/*** 测试分片* @author JiHao*/
public class OrderItem {private Long orderItemId;private Long orderId;private Integer userId;public Long getOrderId() {return orderId;}public void setOrderId(Long orderId) {this.orderId = orderId;}public Integer getUserId() {return userId;}public void setUserId(Integer userId) {this.userId = userId;}public Long getOrderItemId() {return orderItemId;}public void setOrderItemId(Long orderItemId) {this.orderItemId = orderItemId;}}

下面是测试的controller，并没有写Junit测试。

package com.jihao.controller.test;import java.util.List;
import java.util.Map;import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.stereotype.Controller;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.ResponseBody;import com.jihao.dao.TestShardingMapper;
import com.jihao.entity.Order;
import com.jihao.entity.OrderItem;
import com.jihao.result.Result;
import com.jihao.result.ResultUtil;/*** 测试分片* @author JiHao**/
@Controller
@RequestMapping(value = "test")
public class TestShardingController {@Autowiredprivate TestShardingMapper testShardingMapper;/*** 测试添加* @return*/@ResponseBody@GetMapping(value = "/testAdd")public String testAdd(){for (int i = 0; i < 10; i++) {Order order = new Order();
//            order.setUserId(50);
//            order.setUserId(51);
//            order.setUserId(10001);order.setUserId(20001);order.setStatus("INSERT_TEST");int count = testShardingMapper.insert(order);System.out.println(count);long orderId = order.getOrderId();System.out.println(order.getOrderId());OrderItem item = new OrderItem();item.setOrderId(orderId);
//            order.setUserId(50);
//            order.setUserId(51);
//            order.setUserId(10001);order.setUserId(20001);testShardingMapper.insertItem(item);}return "success";}/*** 测试搜索* @return*/@ResponseBody@GetMapping(value = "/testSearch")public Result searchData(){List<Order> list = testShardingMapper.searchOrder();System.out.println(list.size() + " all");List<Order> list1 = testShardingMapper.queryWithIn();System.out.println(list1.size() + " In");List<Order> list2 = testShardingMapper.queryWithEqual();System.out.println(list2.size() + " Equal");List<Order> list3 = testShardingMapper.queryWithBetween();System.out.println(list3.size() + " Between");List<Map<String, Object>> list4 = testShardingMapper.queryUser();System.out.println(list4.size() + " user");return ResultUtil.success(null);}}

这里要重点提出来的是做搜索测试的时候，因为主从库都在我本地服务器上，并没有做主从复制，大家可以根据我上篇博文配置一下就可以顺利操作了，如果没有配置的话从库里是不会有数据的，所以在做完写操作时把主库中的数据手动传输给从库，这样才能读出数据。
这里顺便给出Sharding-Sphere的官方地址http://shardingjdbc.io/index_zh.html，以及demo地址https://github.com/sharding-sphere/sharding-sphere-example（demo里Sharding-Sphere的maven配置我在跑的时候没跑通，需要把版本改成3.0.0.M1就ok了）。