Java整合RabbitMQ实现生产消费(7种通讯方式)

简介: Java整合RabbitMQ实现生产消费(7种通讯方式)

环境说明

  • RabbitMQ环境,参考RabbitMQ环境搭建
  • Java版本:JDK1.8
  • Maven版本:apache-maven-3.6.3
  • 开发工具:IntelliJ IDEA

工程搭建

  1. 创建maven项目
  2. pom.xml文件引入RabbitMQ依赖
    <dependencies>
        <dependency>
            <groupId>com.rabbitmq</groupId>
            <artifactId>amqp-client</artifactId>
            <version>5.16.0</version>
        </dependency>
        <dependency>
            <groupId>junit</groupId>
            <artifactId>junit</artifactId>
            <version>4.13.2</version>
            <scope>compile</scope>
        </dependency>

    </dependencies>

连接RabbitMQ

import com.rabbitmq.client.Connection;
import com.rabbitmq.client.ConnectionFactory;

import java.io.IOException;
import java.util.concurrent.TimeoutException;

public class MQConnections {
   
   

    public static final String RABBITMQ_HOST = "127.0.0.1";
    public static final int RABBITMQ_PORT = 5672;
    public static final String RABBITMQ_USERNAME = "guest";
    public static final String RABBITMQ_PASSWORD = "guest";
    public static final String RABBITMQ_VIRTUAL_HOST = "/";

    /**
     * 构建RabbitMQ连接对象
     *
     * @return
     */
    public static Connection getConnection() throws IOException, TimeoutException {
   
   
        //1.创建Connection工厂
        ConnectionFactory factory = new ConnectionFactory();
        //2.设置Rabbitmq连接信息
        factory.setHost(RABBITMQ_HOST);
        factory.setPort(RABBITMQ_PORT);
        factory.setUsername(RABBITMQ_USERNAME);
        factory.setPassword(RABBITMQ_PASSWORD);
        factory.setVirtualHost(RABBITMQ_VIRTUAL_HOST);
        //3.返回连接对象
        return factory.newConnection();
    }

}

通讯模式

1.简单通讯

即一个生产者可以向一个队列发送消息,一个消费者可以尝试从一个队列接收数据。如下图:
在这里插入图片描述

    public final static String HELLO_QUEUE_NAME = "hello";
    @Test
    public void publish_hello() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(HELLO_QUEUE_NAME, false, false, false, null);
        //4.发布消息
        String msg = "hello,world";
        channel.basicPublish("", HELLO_QUEUE_NAME, null, msg.getBytes());
    }
    @Test
    public void consume_hello() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(HELLO_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("Received '" + message + "'");
        };
        channel.basicConsume(HELLO_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

2.工作队列通讯

与简单通讯一样,当消费能力不足或想要提高吞吐时可添加多个消费者进行处理业务。如下图,队列中的消息会逐条被C1和C2消费。
在这里插入图片描述

    public final static String WORK_QUEUE_NAME = "work";

    @Test
    public void publish_work_queue() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(WORK_QUEUE_NAME, false, false, false, null);
        //4.发布消息
        String msg = "hello,work queue";
        channel.basicPublish("", WORK_QUEUE_NAME, null, msg.getBytes());
    }
        @Test
    public void consume_work_queue1() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(WORK_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("consume1 Received '" + message + "'");
        };
        channel.basicConsume(WORK_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

    @Test
    public void consume_work_queue2() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(WORK_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("consume2 Received '" + message + "'");
        };
        channel.basicConsume(WORK_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

3.发布/订阅通讯

工作队列背后的假设是,每个任务只交付给一个消费者做同一件事。如果要交付给多个消费者做不同的事,需要引入交换机实现一个完整的消息传递模型,这种模式被称为“发布/订阅”。如下图,消息会发布到交换机中,交换机向绑定的队列同时发送消息,最终C1和C2会同时消费此条消息。
在这里插入图片描述

public final static String PUB_EXCHANGE_NAME = "pub-ex";
    public final static String PUB1_QUEUE_NAME = "pub-que1";
    public final static String PUB2_QUEUE_NAME = "pub-que2";

    @Test
    public void publish_pub_sub() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建交换机
        channel.exchangeDeclare(PUB_EXCHANGE_NAME, BuiltinExchangeType.FANOUT);
        //4.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(PUB1_QUEUE_NAME, false, false, false, null);
        channel.queueDeclare(PUB2_QUEUE_NAME, false, false, false, null);
        //5.绑定队列
        channel.queueBind(PUB1_QUEUE_NAME, PUB_EXCHANGE_NAME, "");
        channel.queueBind(PUB2_QUEUE_NAME, PUB_EXCHANGE_NAME, "");
        //6.发布消息
        String msg = "hello,pub/sub";
        channel.basicPublish(PUB_EXCHANGE_NAME, "", null, msg.getBytes());
    }
    @Test
    public void consume_pub_sub1() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(PUB1_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("pub_sub1 Received '" + message + "'");
        };
        channel.basicConsume(PUB1_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

    @Test
    public void consume_pub_sub2() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(PUB2_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("pub_sub2 Received '" + message + "'");
        };
        channel.basicConsume(PUB2_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

4.路由通讯

发布/订阅模式是交换机将一条消息同时路由给多个队列,“路由”模式可以将消息通过交换机指定到某个队列中从而被消费。如下图,交换机将所有类型的日志路由到一个队列中,将error类型的日志路由到另一个队列中。
在这里插入图片描述

    public final static String ROUT_EXCHANGE_NAME = "rout-ex";
    public final static String ROUTALL_QUEUE_NAME = "rout-queall";
    public final static String ROUTONE_QUEUE_NAME = "rout-queone";

    @Test
    public void publish_routing() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建交换机
        channel.exchangeDeclare(ROUT_EXCHANGE_NAME, BuiltinExchangeType.DIRECT);
        //4.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(ROUTALL_QUEUE_NAME, false, false, false, null);
        channel.queueDeclare(ROUTONE_QUEUE_NAME, false, false, false, null);
        //5.绑定队列
        channel.queueBind(ROUTALL_QUEUE_NAME, ROUT_EXCHANGE_NAME, "all");
        channel.queueBind(ROUTONE_QUEUE_NAME, ROUT_EXCHANGE_NAME, "one");
        //6.发布消息
        String msg1 = "hello,1-all";
        String msg2 = "hello,2-all";
        String msg3 = "hello,1-one";
        channel.basicPublish(ROUT_EXCHANGE_NAME, "all", null, msg1.getBytes());
        channel.basicPublish(ROUT_EXCHANGE_NAME, "all", null, msg2.getBytes());
        channel.basicPublish(ROUT_EXCHANGE_NAME, "one", null, msg3.getBytes());
    }
    @Test
    public void consume_routing_all() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(ROUTALL_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("consume_routing_all Received '" + message + "'");
        };
        channel.basicConsume(ROUTALL_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

    @Test
    public void consume_routing_one() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(ROUTONE_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("consume_routing_one Received '" + message + "'");
        };
        channel.basicConsume(ROUTONE_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

5.主题通讯

“路由”模式仍然有局限性——它不能基于多个标准进行路由。主题可以带来很大的灵活性,发送到主题交换的消息不能有任意的routing_key,它必须是一个用点分隔的单词列表,routing_key有两种重要的特殊情况:

  • *只能代替一个词。
  • 可以替换零个或多个单词。

    在这里插入图片描述

      public final static String TOPIC_EXCHANGE_NAME = "topic-ex";
      public final static String TOPICALL_QUEUE_NAME = "topic-queall";
      public final static String TOPICONE_QUEUE_NAME = "topic-queone";
    
      @Test
      public void publish_topic() throws IOException, TimeoutException {
         
         
          //1.获取连接对象
          Connection connection = MQConnections.getConnection();
          //2.构建Channl
          Channel channel = connection.createChannel();
          //3.构建交换机
          channel.exchangeDeclare(TOPIC_EXCHANGE_NAME, BuiltinExchangeType.TOPIC);
          //4.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
          channel.queueDeclare(TOPICALL_QUEUE_NAME, false, false, false, null);
          channel.queueDeclare(TOPICONE_QUEUE_NAME, false, false, false, null);
          //5.绑定队列
          channel.queueBind(TOPICALL_QUEUE_NAME, TOPIC_EXCHANGE_NAME, "*.all.*");
          channel.queueBind(TOPICONE_QUEUE_NAME, TOPIC_EXCHANGE_NAME, "#.one");
          //6.发布消息
          String msg1 = "hello.all.world";
          String msg2 = "hello.world.one";
          channel.basicPublish(TOPIC_EXCHANGE_NAME, "hello.all.world", null, msg1.getBytes());
          channel.basicPublish(TOPIC_EXCHANGE_NAME, "hello.world.one", null, msg2.getBytes());
      }
          @Test
      public void consume_topic_all() throws IOException, TimeoutException {
         
         
          //1.获取连接对象
          Connection connection = MQConnections.getConnection();
          //2.构建Channl
          Channel channel = connection.createChannel();
          //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
          channel.queueDeclare(TOPICALL_QUEUE_NAME, false, false, false, null);
          //4.监听消息
          DeliverCallback deliverCallback = (consumerTag, delivery) -> {
         
         
              String message = new String(delivery.getBody(), "UTF-8");
              System.out.println("consume_topic_all Received '" + message + "'");
          };
          channel.basicConsume(TOPICALL_QUEUE_NAME, true, deliverCallback, consumerTag -> {
         
         
          });
          System.in.read();
      }
    
      @Test
      public void consume_topic_one() throws IOException, TimeoutException {
         
         
          //1.获取连接对象
          Connection connection = MQConnections.getConnection();
          //2.构建Channl
          Channel channel = connection.createChannel();
          //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
          channel.queueDeclare(TOPICONE_QUEUE_NAME, false, false, false, null);
          //4.监听消息
          DeliverCallback deliverCallback = (consumerTag, delivery) -> {
         
         
              String message = new String(delivery.getBody(), "UTF-8");
              System.out.println("consume_topic_one Received '" + message + "'");
          };
          channel.basicConsume(TOPICONE_QUEUE_NAME, true, deliverCallback, consumerTag -> {
         
         
          });
          System.in.read();
      }
    

    6.RPC通讯

    RabbitMQ作为消息中间件可以达到应用解耦效果,如果想达到RPC远程调用同步返回结果,RabbitMQ同样支持,其原理如下:

  • 发布者发送消息时指定一个回调队列和唯一id
  • 消费者处理完成后将结果发送到回调队列中
  • 发布者按照唯一id接收消息并处理

如下图
在这里插入图片描述

    public final static String RPC_QUEUE_NAME = "rpc-que";
    public final static String RPCCALLBACK_QUEUE_NAME = "rpc-callback-que";

    @Test
    public void publish_rpc() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(RPC_QUEUE_NAME, false, false, false, null);
        // replyTo回调队列
        channel.queueDeclare(RPCCALLBACK_QUEUE_NAME, false, false, false, null);
        //4.发布消息
        String msg = "hello rpc";
        String correlationId = UUID.randomUUID().toString();
        /*AMQP 协议预先定义了一组与消息一起使用的14个属性。除了以下属性外,大多数属性很少使用:
        deliveryMode:将消息标记为持久(值为2)或瞬时(任何其他值)。
        contentType:用于描述编码的mime类型。例如,对于常用的JSON编码,最好将此属性设置为:application/JSON。
        replyTo:通常用于命名回调队列。
        correlationId:用于将RPC响应与请求关联。*/
        AMQP.BasicProperties basicProperties = new AMQP.BasicProperties.Builder().replyTo(RPCCALLBACK_QUEUE_NAME).correlationId(correlationId).build();
        //5.回调响应结果
        channel.basicPublish("", RPC_QUEUE_NAME, basicProperties, msg.getBytes());
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            String recid = delivery.getProperties().getCorrelationId();
            if (correlationId.equalsIgnoreCase(recid)) System.out.println("rpc-callback-que   '" + message + "'");
        };
        channel.basicConsume(RPCCALLBACK_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }
        @Test
    public void consume_rpc() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(RPC_QUEUE_NAME, false, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("consume_rpc Received '" + message + "'");
            String correlationId = delivery.getProperties().getCorrelationId();
            AMQP.BasicProperties basicProperties = new AMQP.BasicProperties.Builder().correlationId(correlationId).build();
            String replyTo = delivery.getProperties().getReplyTo();
            String callbackmsg = "rpc callback";
            channel.basicPublish("", replyTo, basicProperties, callbackmsg.getBytes());
        };
        channel.basicConsume(RPC_QUEUE_NAME, true, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

7.Publisher确认通讯

Publisher确认是RabbitMQ扩展以实现可靠发布。当在通道上启用发布者确认时,代理将异步确认客户端发布的消息,这意味着它们已在服务器端得到处理。

    public final static String CONFIRM_EXCHANGE_NAME = "confirm-ex";
    public final static String CONFIRM_QUEUE_NAME = "confirm-que";

    @Test
    public void publish_confirm() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.开启确认选项
        channel.confirmSelect();
        //4.构建交换机
        channel.exchangeDeclare(CONFIRM_EXCHANGE_NAME, BuiltinExchangeType.DIRECT);
        //5.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(CONFIRM_QUEUE_NAME, true, false, false, null);
        //6.绑定队列
        String right_routing_key = "confirm";
        String error_routing_key = "confirm_err";
        channel.queueBind(CONFIRM_QUEUE_NAME, CONFIRM_EXCHANGE_NAME, right_routing_key);
        //7.消息到达交换机确认监听
        channel.addConfirmListener((sequenceNumber, multiple) -> {
   
   
            System.out.println("消息成功发送到交换机");
        }, (sequenceNumber, multiple) -> {
   
   
            System.err.println("消息未发送到交换机,补偿操作。");
        });
        //8.消息到达队列确认监听
        channel.addReturnListener((replyCode, replyText, exchange, routingKey, basicProperties, body) -> {
   
   
            System.err.format("消息 %s 未路由到指定队列: %s, replyText: %s,replyCode: %d%n", body, routingKey, replyText, replyCode);
        });
        //设置消息持久化
        AMQP.BasicProperties basicProperties = new AMQP.BasicProperties.Builder().deliveryMode(2).build();
        //7.发布消息
        String msg = "hello confirm";
        channel.basicPublish(CONFIRM_EXCHANGE_NAME, error_routing_key,true, basicProperties, msg.getBytes());
        System.in.read();
    }

    @Test
    public void consume_ack() throws IOException, TimeoutException {
   
   
        //1.获取连接对象
        Connection connection = MQConnections.getConnection();
        //2.构建Channl
        Channel channel = connection.createChannel();
        //3.构建队列,queueDeclare("队列名称","是否持久化队列","是否只允许一个队列消费","长时间未使用是否删除","其他参数")
        channel.queueDeclare(CONFIRM_QUEUE_NAME, true, false, false, null);
        //4.监听消息
        DeliverCallback deliverCallback = (consumerTag, delivery) -> {
   
   
            String message = new String(delivery.getBody(), "UTF-8");
            System.out.println("consume_routing_one Received '" + message + "'");
            //消息处理后手动ACK
            channel.basicAck(delivery.getEnvelope().getDeliveryTag(),false);
        };
        // ack为false
        channel.basicConsume(CONFIRM_QUEUE_NAME, false, deliverCallback, consumerTag -> {
   
   
        });
        System.in.read();
    }

代码仓库

https://gitee.com/codeWBG/learn_rabbitmq

相关实践学习
消息队列RocketMQ版:基础消息收发功能体验
本实验场景介绍消息队列RocketMQ版的基础消息收发功能,涵盖实例创建、Topic、Group资源创建以及消息收发体验等基础功能模块。
消息队列 MNS 入门课程
1、消息队列MNS简介 本节课介绍消息队列的MNS的基础概念 2、消息队列MNS特性 本节课介绍消息队列的MNS的主要特性 3、MNS的最佳实践及场景应用 本节课介绍消息队列的MNS的最佳实践及场景应用案例 4、手把手系列:消息队列MNS实操讲 本节课介绍消息队列的MNS的实际操作演示 5、动手实验:基于MNS,0基础轻松构建 Web Client 本节课带您一起基于MNS,0基础轻松构建 Web Client
相关文章
|
30天前
|
存储 消息中间件 安全
JUC组件实战:实现RRPC(Java与硬件通过MQTT的同步通信)
【10月更文挑战第9天】本文介绍了如何利用JUC组件实现Java服务与硬件通过MQTT的同步通信(RRPC)。通过模拟MQTT通信流程,使用`LinkedBlockingQueue`作为消息队列,详细讲解了消息发送、接收及响应的同步处理机制,包括任务超时处理和内存泄漏的预防措施。文中还提供了具体的类设计和方法实现,帮助理解同步通信的内部工作原理。
JUC组件实战:实现RRPC(Java与硬件通过MQTT的同步通信)
|
1月前
|
消息中间件 存储 JSON
rabbitmq基础教程(ui,java,springamqp)
本文提供了RabbitMQ的基础教程,包括如何使用UI创建队列和交换机、Java代码操作RabbitMQ、Spring AMQP进行消息发送和接收,以及如何使用不同的交换机类型(fanout、direct、topic)进行消息路由。
22 0
rabbitmq基础教程(ui,java,springamqp)
|
1月前
|
消息中间件 前端开发 Java
java高并发场景RabbitMQ的使用
java高并发场景RabbitMQ的使用
81 0
|
2月前
|
消息中间件 缓存 Java
RocketMQ的JAVA落地实战
RocketMQ作为一款高性能、高可靠、高实时、分布式特点的消息中间件,其核心作用主要体现在异步处理、削峰填谷以及系统解耦三个方面。
168 0
|
3月前
|
消息中间件 负载均衡 Java
Java微服务通讯方式有哪些?
【8月更文挑战第18天】Java微服务通讯方式有哪些?
48 1
|
3月前
|
Java
MQTT(EMQX) - Java 调用 MQTT Demo 代码
MQTT(EMQX) - Java 调用 MQTT Demo 代码
151 0
MQTT(EMQX) - Java 调用 MQTT Demo 代码
|
3月前
|
设计模式 安全 Java
Java多线程通讯
这些机制都是多线程通信的关键,使用它们可以在Java中实现高效、安全的线程协作。在实现多线程通信时,需注意死锁、饥饿、竞态条件等多线程问题。确保对多线程的理解深入和正确的设计模式应用将会是编写稳健多线程程序的基础。
14 0
|
4月前
|
消息中间件 Java Maven
如何在Java中使用RabbitMQ
如何在Java中使用RabbitMQ
|
4月前
|
存储 设计模式 监控
Java面试题:如何在不牺牲性能的前提下,实现一个线程安全的单例模式?如何在生产者-消费者模式中平衡生产和消费的速度?Java内存模型规定了变量在内存中的存储和线程间的交互规则
Java面试题:如何在不牺牲性能的前提下,实现一个线程安全的单例模式?如何在生产者-消费者模式中平衡生产和消费的速度?Java内存模型规定了变量在内存中的存储和线程间的交互规则
48 0
|
4月前
|
消息中间件 负载均衡 Java
JAVA面试之MQ
JAVA面试之MQ
69 0