EventProcessor与WorkPool用法--可处理多消费者

 

单一的生产者，消费者有多个，使用WorkerPool来管理多个消费者；

 

RingBuffer在生产Sequencer中记录一个cursor，追踪生产者生产到的最新位置，通过WorkSequence和sequence记录整个workpool消费的位置和每个WorkProcessor消费到位置，来协调生产和消费程序

 

1、定义事件

package com.ljq.disruptor;

import java.io.Serializable;

/**
 * 交易事件数据
 * 
 * @author Administrator
 *
 */
@SuppressWarnings("serial")
public class TradeEvent implements Serializable {
    private String id; // 订单ID
    private String name;
    private double price; // 金额

    public TradeEvent() {
    }
    
    public TradeEvent(String id) {
        super();
        this.id = id;
    }

    public String getId() {
        return id;
    }

    public void setId(String id) {
        this.id = id;
    }

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }

    public double getPrice() {
        return price;
    }

    public void setPrice(double price) {
        this.price = price;
    }

    @Override
    public String toString() {
        return "Trade [id=" + id + ", name=" + name + ", price=" + price + "]";
    }

}

 

2、TradeEvent事件消费者

package com.ljq.disruptor;

import com.lmax.disruptor.EventHandler;
import com.lmax.disruptor.WorkHandler;

public class TradeEventHandler implements EventHandler<TradeEvent>, WorkHandler<TradeEvent> {
    @Override
    public void onEvent(TradeEvent event, long sequence, boolean endOfBatch) throws Exception {
        this.onEvent(event);
    }

    /**
     * WorkProcessor多线程排队领event然后再执行，不同线程执行不同的event。但是多了个排队领event的过程，这个是为了减少对生产者队列查询的压力
     */
    @Override
    public void onEvent(TradeEvent event) throws Exception {
        // 具体的消费逻辑
        System.out.println("consumer:" + Thread.currentThread().getName() + " Event: value=" + event);
    }
}

 

3、EventProcessor消费者-生产者启动类

package com.ljq.disruptor;

import java.util.UUID;
import java.util.concurrent.Callable;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.Future;

import com.lmax.disruptor.BatchEventProcessor;
import com.lmax.disruptor.EventFactory;
import com.lmax.disruptor.RingBuffer;
import com.lmax.disruptor.SequenceBarrier;
import com.lmax.disruptor.YieldingWaitStrategy;

public class EventProcessorMain {
    public static void main(String[] args) throws Exception {  
        long beginTime = System.currentTimeMillis();
        
        // 指定 ring buffer字节大小，必需为2的N次方(能将求模运算转为位运算提高效率 )，否则影响性能
        int bufferSize = 1024;
        //固定线程数
        int nThreads = 4;
        
        EventFactory<TradeEvent> eventFactory = new EventFactory<TradeEvent>() {  
            @Override  
            public TradeEvent newInstance() {  
                return new TradeEvent(UUID.randomUUID().toString());
            }  
        };
        
        //RingBuffer. createSingleProducer创建一个单生产者的RingBuffer
        //第一个参数叫EventFactory，从名字上理解就是“事件工厂”，其实它的职责就是产生数据填充RingBuffer的区块。 
        //第二个参数是RingBuffer的大小，它必须是2的整数倍，目的是为了将求模运算转为&运算提高效率
        //第三个参数是RingBuffer的生产在没有可用区块的时候(可能是消费者太慢了)的等待策略 
        final RingBuffer<TradeEvent> ringBuffer = RingBuffer.createSingleProducer(eventFactory, bufferSize, new YieldingWaitStrategy());  
        
        //SequenceBarrier, 协调消费者与生产者, 消费者链的先后顺序. 阻塞后面的消费者(没有Event可消费时)
        SequenceBarrier sequenceBarrier = ringBuffer.newBarrier();  
          
        //创建消费者事件处理器， 多线程并发执行，不同线程执行不同的event 
        BatchEventProcessor<TradeEvent> transProcessor = new BatchEventProcessor<TradeEvent>(ringBuffer, sequenceBarrier, new TradeEventHandler());  
        //把消费者的消费进度情况注册给RingBuffer结构(生产者)，如果只有一个消费者的情况可以省略 
        ringBuffer.addGatingSequences(transProcessor.getSequence());  
          
        //创建一个可重用固定线程数的线程池，以共享的无界队列方式来运行这些线程
        ExecutorService executors = Executors.newFixedThreadPool(nThreads);  
        //把消费者提交到线程池，说明EventProcessor实现了callable接口  
        executors.submit(transProcessor);  
        
        // 生产者，这里新建线程不是必要的
        Future<?> future= executors.submit(new Callable<Void>() {  
            @Override  
            public Void call() throws Exception {  
                long seq;  
                for (int i = 0; i < 100000; i++) {
                    seq = ringBuffer.next();
                    ringBuffer.get(seq).setPrice(i);
                    ringBuffer.publish(seq);
                } 
                return null;  
            }  
        }); 
        future.get();//等待生产者结束  
        
        Thread.sleep(1000); //等上1秒，等消费都处理完成
        transProcessor.halt(); //通知事件(或者说消息)处理器 可以结束了（并不是马上结束!!!） 
        executors.shutdown(); 
        
        System.out.println(String.format("总共耗时%s毫秒", (System.currentTimeMillis() - beginTime)));

    }  
}

 

4、WorkerPool消费者-生产者启动类

package com.ljq.disruptor;

import java.util.UUID;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;

import com.lmax.disruptor.EventFactory;
import com.lmax.disruptor.IgnoreExceptionHandler;
import com.lmax.disruptor.RingBuffer;
import com.lmax.disruptor.SequenceBarrier;
import com.lmax.disruptor.WorkerPool;

public class WorkPoolMain {
    public static void main(String[] args) throws InterruptedException {
        // 指定 ring buffer字节大小，必需为2的N次方(能将求模运算转为位运算提高效率 )，否则影响性能
        int bufferSize = 1024;
        //固定线程数
        int nThreads = 4;

        //RingBuffer. createSingleProducer创建一个单生产者的RingBuffer
        RingBuffer<TradeEvent> ringBuffer = RingBuffer.createSingleProducer(new EventFactory<TradeEvent>() {
            public TradeEvent newInstance() {
                return new TradeEvent(UUID.randomUUID().toString());
            }
        }, bufferSize);

        SequenceBarrier sequenceBarrier = ringBuffer.newBarrier();

        WorkerPool<TradeEvent> workerPool = new WorkerPool<TradeEvent>(ringBuffer, sequenceBarrier,
                new IgnoreExceptionHandler(), new TradeEventHandler());

        //创建一个可重用固定线程数的线程池，以共享的无界队列方式来运行这些线程
        ExecutorService executors = Executors.newFixedThreadPool(nThreads);
        workerPool.start(executors);

        // 生产10个数据
        for (int i = 0; i < 80000; i++) {
            long seq = ringBuffer.next();
            ringBuffer.get(seq).setPrice(i);
            ringBuffer.publish(seq);
        }

        Thread.sleep(1000); //等上1秒，等消费都处理完成
        workerPool.halt(); //通知事件(或者说消息)处理器 可以结束了（并不是马上结束!!!） 
        executors.shutdown(); 
    }
}