1. 多数据源实现的核心挑战与解决方案在业务系统开发中多数据源需求通常源于以下场景需要同时连接多个不同类型的数据库如MySQLPostgreSQL、读写分离架构、分库分表实现、多租户SaaS系统等。传统单数据源配置无法满足这些复杂场景我们需要一套灵活的数据源路由机制。实现多数据源的核心难点在于如何管理多个数据源的生命周期如何在运行时动态切换数据源如何保证事务上下文的一致性如何避免连接泄漏等问题Spring框架提供的AbstractRoutingDataSource正是为解决这些问题而设计的抽象类。它本质上是一个数据源的路由器内部维护了多个实际数据源的映射关系根据指定的key动态返回对应的真实数据源。2. 基于AbstractRoutingDataSource的实现原理2.1 核心架构设计AbstractRoutingDataSource的工作流程可以分为三个关键环节初始化阶段将所有目标数据源注册到Map结构中路由决策阶段通过determineCurrentLookupKey()方法获取当前数据源key连接获取阶段根据key从Map中获取真实数据源并返回连接public abstract class AbstractRoutingDataSource extends AbstractDataSource { private MapObject, DataSource targetDataSources; private Object defaultTargetDataSource; protected abstract Object determineCurrentLookupKey(); public Connection getConnection() throws SQLException { return determineTargetDataSource().getConnection(); } }2.2 线程上下文绑定机制数据源切换的核心在于如何存储和传递路由key。ThreadLocal是解决这个问题的理想选择因为它具有以下特性线程隔离每个线程独立存储自己的数据源key隐式传递无需显式传递参数方法调用链自动继承上下文生命周期可控请求结束时可以清理ThreadLocal防止内存泄漏典型实现方案public class DataSourceContextHolder { private static final ThreadLocalString contextHolder new ThreadLocal(); public static void setDataSource(String name) { contextHolder.set(name); } public static String getDataSource() { return contextHolder.get(); } public static void clear() { contextHolder.remove(); } }3. 完整实现步骤详解3.1 基础环境配置首先在Spring Boot配置文件中定义多个数据源spring: datasource: master: url: jdbc:mysql://localhost:3306/master_db username: root password: 123456 driver-class-name: com.mysql.cj.jdbc.Driver slave: url: jdbc:mysql://localhost:3306/slave_db username: root password: 123456 driver-class-name: com.mysql.cj.jdbc.Driver3.2 动态数据源配置类创建继承AbstractRoutingDataSource的DynamicDataSourcepublic class DynamicDataSource extends AbstractRoutingDataSource { Override protected Object determineCurrentLookupKey() { return DataSourceContextHolder.getDataSource(); } }配置数据源BeanConfiguration public class DataSourceConfig { Bean ConfigurationProperties(spring.datasource.master) public DataSource masterDataSource() { return DataSourceBuilder.create().build(); } Bean ConfigurationProperties(spring.datasource.slave) public DataSource slaveDataSource() { return DataSourceBuilder.create().build(); } Bean public DataSource dynamicDataSource( Qualifier(masterDataSource) DataSource master, Qualifier(slaveDataSource) DataSource slave) { DynamicDataSource dataSource new DynamicDataSource(); MapObject, Object targetDataSources new HashMap(); targetDataSources.put(master, master); targetDataSources.put(slave, slave); dataSource.setTargetDataSources(targetDataSources); dataSource.setDefaultTargetDataSource(master); return dataSource; } }3.3 数据源切换注解实现创建自定义注解用于标记需要切换数据源的方法Target({ElementType.METHOD, ElementType.TYPE}) Retention(RetentionPolicy.RUNTIME) Documented public interface DataSource { String value() default master; }通过AOP实现注解拦截Aspect Component public class DataSourceAspect { Before(annotation(dataSource)) public void beforeSwitchDataSource(JoinPoint point, DataSource dataSource) { String dsName dataSource.value(); DataSourceContextHolder.setDataSource(dsName); } After(annotation(dataSource)) public void afterSwitchDataSource(JoinPoint point, DataSource dataSource) { DataSourceContextHolder.clear(); } }4. 高级特性与生产级优化4.1 事务管理增强多数据源环境下的事务管理需要特别注意分布式事务问题跨数据源的操作无法通过本地事务保证一致性事务传播行为确保事务方法内部的数据源切换不会导致意外行为解决方案对于强一致性要求高的场景引入Seata等分布式事务框架在事务方法开始前就确定数据源避免事务中途切换添加事务拦截器确保数据源一致性Around(annotation(transactional)) public Object aroundTransactional(ProceedingJoinPoint joinPoint, Transactional transactional) throws Throwable { String dsName determineDataSource(joinPoint); DataSourceContextHolder.setDataSource(dsName); try { return joinPoint.proceed(); } finally { DataSourceContextHolder.clear(); } }4.2 连接池优化配置生产环境建议使用Druid连接池并针对多数据源优化Bean ConfigurationProperties(spring.datasource.master) public DataSource masterDataSource() { return DruidDataSourceBuilder.create().build(); } // Druid监控配置 Bean public ServletRegistrationBeanStatViewServlet druidServlet() { ServletRegistrationBeanStatViewServlet reg new ServletRegistrationBean(); reg.setServlet(new StatViewServlet()); reg.addUrlMappings(/druid/*); reg.addInitParameter(loginUsername, admin); reg.addInitParameter(loginPassword, admin); return reg; }关键参数建议initialSize: 5-10 (根据并发量调整)maxActive: 20-100minIdle: 5-10maxWait: 60000mstimeBetweenEvictionRunsMillis: 600005. 常见问题排查指南5.1 典型错误与解决方案问题现象可能原因解决方案No qualifying bean of type javax.sql.DataSource available未正确配置主数据源检查Primary注解或显式指定数据源bean名称Failed to configure a DataSource: url attribute is not specified配置属性不匹配检查spring.datasource.xxx前缀是否与代码一致Connection is closed unexpectedlyThreadLocal未及时清理确保在finally块中调用clear()Transaction rolled back because it has been marked as rollback-only跨数据源事务问题避免在事务中切换数据源或引入分布式事务5.2 性能优化建议连接泄漏检测定期检查活跃连接数添加连接回收机制路由缓存对频繁访问的路由key做本地缓存预热连接池应用启动时预先建立部分连接监控告警实现数据源健康检查接口public class DataSourceHealthIndicator implements HealthIndicator { private final MapString, DataSource dataSources; Override public Health health() { MapString, Object details new HashMap(); dataSources.forEach((name, ds) - { try (Connection conn ds.getConnection()) { details.put(name, UP); } catch (Exception e) { details.put(name, DOWN - e.getMessage()); } }); return Health.up().withDetails(details).build(); } }6. 扩展应用场景6.1 多租户系统实现基于多数据源架构可以实现SaaS多租户方案public class TenantDataSourceInterceptor implements HandlerInterceptor { Override public boolean preHandle(HttpServletRequest request, HttpServletResponse response, Object handler) { String tenantId request.getHeader(X-Tenant-ID); if (StringUtils.isNotBlank(tenantId)) { DataSourceContextHolder.setDataSource(tenant_ tenantId); } return true; } Override public void afterCompletion(HttpServletRequest request, HttpServletResponse response, Object handler, Exception ex) { DataSourceContextHolder.clear(); } }6.2 读写分离扩展通过AOP自动路由读操作到从库Pointcut(execution(* com..repository.*.find*(..)) || execution(* com..repository.*.get*(..)) || execution(* com..repository.*.query*(..))) public void readOperation() {} Before(readOperation()) public void beforeReadOperation() { if (DataSourceContextHolder.getDataSource() null) { DataSourceContextHolder.setDataSource(slave); } }在实际项目中我们还需要考虑从库延迟、负载均衡等问题。可以通过以下方式增强添加从库健康检查实现简单的负载均衡策略对于强一致性要求的读操作支持强制走主库Target(ElementType.METHOD) Retention(RetentionPolicy.RUNTIME) public interface MasterRoute { // 强制走主库的标记注解 } Before(annotation(masterRoute)) public void beforeMasterRoute() { DataSourceContextHolder.setDataSource(master); }