Disruptor并发框架简介
- Disruptor是一个开源的并发框架,能够在无锁的情况下实现网络的Queue并发操作。
- Disruptor是一个高性能的异步处理框架,或者可以认为是最快的消息框架(轻量级JMS),也可以认为是一个观察者模式的实现,或者事件监听模式的实现。
- 参考系列文档:http://ifeve.com/disruptor-getting-started/
Disruptor Hello World
在Disruptor中,进行开发时需要引入相应的jar包。我们实现Hello World需要如下几个步骤:
- 建立一个Event类。
- 建立一个工厂Event类,用于创建Event类实例对象。
- 需要有一个事件监听事件类,用于处理数据(Event类)。
- 我们需要进行测试代码编写。实例化Disruptor实例,配置一系列参数。然后我们对Disruptor实例绑定监听事件类,接收并处理数据。
- 在Disruptor中,真正存储数据的核心叫做RingBuffer,我们通过Disruptor实例来拿到它,然后把数据生产出来,把数据加入到RingBuffer的实例对象中即可。
public class LongEvent {
private long value;
public long getValue() {
return value;
}
public void setValue(long value) {
this.value = value;
}
}
/**
* 需要让disruptor为我们创建事件,我们同时还声明一个EventFactory来实例化Event对象
* */
public class LongEventFactory implements EventFactory{
@Override
public Object newInstance() {
return new LongEvent();
}
}
/**
* 事件消费者:
* 也就是一个事件处理器,这个事件处理器简单地把事件中存储的数据打印到终端
* */
public class LongEventHandler implements EventHandler<LongEvent> {
@Override
public void onEvent(LongEvent arg0, long arg1, boolean arg2) throws Exception {
System.out.println(arg0.getValue());
}
}
public class LongEventMain {
public static void main(String[] args) {
//创建线程池
ExecutorService executor = Executors.newCachedThreadPool();
//创建工程
LongEventFactory factory = new LongEventFactory();
//创建bufferSize,也就是RingBuffer大小,必须是2的N次方
int ringBufferSize = 1024 * 1024;
/**
BlockingWaitStrategy 是最低效的策略,但其对CPU的消耗最小并且在各种不同部署环境中能提供更加一致的性能表现
WaitStrategy BLOCKING_WAIT = new BlockingWaitStrategy();
SleepingWaitStrategy 的性能表现跟BlockingWaitStrategy差不多,对CPU的消耗也类似,但其对生产者线程的影响最小,适合用于异步日志类似的场景
WaitStrategy SLEEPING_WAIT = new SleepingWaitStrategy();
YieldingWaitStrategy 的性能是最好的,适合用于低延迟的系统。在要求极高性能且事件处理线数小于CPU逻辑核心数的场景中,推荐使用此策略;例如,CPU开启超线程的特性
WaitStrategy YIELDING_WAIT = new YieldingWaitStrategy();
*/
/* 创建disruptor
* 1.第一个参数为工厂类对象,用于创建一个个的LongEvent,LongEvent是实际的消费数据。
* 2.第二个参数为缓存区大小
* 3.第三个参数为线程池,进行Disruptor内部的数据接收处理调度
* 4.第四个参数ProducerType.SINGLE(表示一个生产者) 和 ProducerType.MULTI(多个生产者)
* 5.第五个参数是一种策略,就是生产和消费的策略
* */
Disruptor<LongEvent> disruptor = new Disruptor<>(factory, ringBufferSize, executor, ProducerType.SINGLE, new YieldingWaitStrategy());
//连接消费事件方法
disruptor.handleEventsWith(new LongEventHandler());
//启动
disruptor.start();
//Disruptor的事件发布过程是一个两阶段提交过程
//使用该方法获得具体存放数据的容器RingBuffer(环形结构)
RingBuffer<LongEvent> ringBuffer = disruptor.getRingBuffer();
LongEventProducer producer = new LongEventProducer(ringBuffer);
ByteBuffer byteBuffer = ByteBuffer.allocate(8);
for(long input = 0; input < 100; input++){
byteBuffer.putLong(0,input);
producer.onData(byteBuffer);
}
disruptor.shutdown(); //关闭disruptor,方法会阻塞,直到所有的时间得到处理
executor.shutdown(); //关闭executor,disruptor不会自动关闭executor
}
}
public class LongEventProducer {
private final RingBuffer<LongEvent> ringBuffer;
public LongEventProducer(RingBuffer<LongEvent> ringBuffer){
this.ringBuffer = ringBuffer;
}
/**
* onData用来发布事件,每调用一次就发布一次事件
* 他的参数会用该事件传递给消费者
* */
public void onData(ByteBuffer input){
//1.把ringBuffer看做一个事件队列,那么next就是得到下一个事件槽
long sequence = ringBuffer.next();
try{
//2.用上面的索引取出一个空的事件用于填充(获取该序号对应的事件对象)
LongEvent event = ringBuffer.get(sequence);
//3.获取要通过事件传递的业务数据
event.setValue(input.getLong(0));
} finally {
//4.发布事件
//注意,最后的ringBuffer.publish 方法必须包含在finally中以确保必须得到调用
//如果某个请求的sequence未被提交,则对应的消费者获取不了数据
ringBuffer.publish(sequence);
}
}
}
Disruptor术语说明
- RingBuffer:
被看做Disruptor最主要组件,然而从3.0开始RingBuffer仅仅负责存储和更新在Disruptor中流通的数据。对一些特殊的使用场景能够被用户(其他数据结构)完全替代。 - Sequence:
Disruptor使用Sequence来表示一个特殊组件处理的序号。和Disruptor一样,每个消费者(EventProcessor)都维持着一个Sequence。大部分的并发代码依赖这些Sequence值得运转,因此Sequence支持多种当前为AtomicLong类的特性。 - Sequencer:
这是Disruptor真正的核心。实现了这个接口的两种生产者(单生产者和多生产者)均实现了所有的并发算法,为了在生产者和消费者之间进行准确快速的数据传递。 - SequenceBarrier:
有Sequence生成,并且包含了已经发布的Sequence的引用,这些Sequence源于Sequenceer和一些独立的消费者的Sequence。它包含了决定是否有供消费者来消费者的Event的逻辑。 - WaitStrategy:
决定了一个消费者将如何等待生产者将Event置于Disruptor。 - Event:
从生产者到消费者过程中所处理的数据单元。Disruptor中没有代码表示Event,因此它完全是由用户定义。 - EventProcessor:
主要时间循环,处理Disruptor中的Event,并且拥有消费者的Sequence。他有一个实现类是BatchEventProcessor,包含了event loop有效的实现,并且将回调到一个EventHandler接口的实现对象。 - EventHandler:
由用户实现并且代表了Disruptor中的一个消费者的接口。 - Producer:
由用户实现,它调用RingBuffer来插入事件(Event),在Disruptor中没有相应的实现代码,由用户实现。 - WorkProcessor:
确保每个sequence只被一个processor消费,在同一个WorkPool中的处理多个WorkProcessor不会消费同样的sequence。 - WorkerPool:
一个WorkProcessor池,其中WorkProcessor将消费Sequence,所以任务可以在实现WorkHandler接口的worker之间移交。 - LifecycleAware:
当BatchEventProcessor启动和停止时,实现这个接口用于接收通知。
Disruptor印象
初看Disruptor,给人的印象就是RingBuffer是其和兴,生产者向RingBuffer中写入元素,消费者从RingBuffer中消费元素。
理解RingBuffer
- RingBuffer到底是什么?
它是一个环(首尾相接的环),可以把它用作在不同上下文(线程)间传递数据的buffer。RingBuffer拥有一个序号,这个序号(sequence)指向数组中下一个可用元素。随着你不停地填充这个buffer(可能会有相应的读取),这个序号一直增长,直到绕过这个环。要找到数组中当前序号指向的元素,可以通过mod操作:sequence mod array.length = array.index(取模操作)。如果槽的个数是2的N次方更有利于基于二进制的计算机计算。
场景使用
在HelloWorld的实例中,我们创建Disruptor实例,然后调用getRingBuffer方法区获取RingBuffer,其实很多时候,我们可以直接使用RingBuffer,以及其他的API操作。
- 使用EventProcessor消息处理器:
public class Main1 {
public static void main(String[] args) throws Exception {
int BUFFER_SIZE=1024;
int THREAD_NUMBERS=4;
/*
* 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的指数倍 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的生产都在没有可用区块的时候(可能是消费者(或者说是事件处理器) 太慢了)的等待策略
*/
final RingBuffer<Trade> ringBuffer = RingBuffer.createSingleProducer(new EventFactory<Trade>() {
@Override
public Trade newInstance() {
return new Trade();
}
}, BUFFER_SIZE, new YieldingWaitStrategy());
//创建线程池
ExecutorService executors = Executors.newFixedThreadPool(THREAD_NUMBERS);
//创建SequenceBarrier
SequenceBarrier sequenceBarrier = ringBuffer.newBarrier();
//创建消息处理器
BatchEventProcessor<Trade> transProcessor = new BatchEventProcessor<Trade>(
ringBuffer, sequenceBarrier, new TradeHandler());
//这一步的目的就是把消费者的位置信息引用注入到生产者 如果只有一个消费者的情况可以省略
ringBuffer.addGatingSequences(transProcessor.getSequence());
//把消息处理器提交到线程池
executors.submit(transProcessor);
//如果存在多个消费者 那重复执行上面3行代码 把TradeHandler换成其它消费者类
Future<?> future= executors.submit(new Callable<Void>() {
@Override
public Void call() throws Exception {
long seq;
for(int i=0;i<10;i++){
seq = ringBuffer.next();//占个坑 --ringBuffer一个可用区块
ringBuffer.get(seq).setPrice(Math.random()*9999);//给这个区块放入 数据
ringBuffer.publish(seq);//发布这个区块的数据使handler(consumer)可见
}
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();//终止线程
}
}
public class TradeHandler implements EventHandler<Trade>, WorkHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
this.onEvent(event);
}
@Override
public void onEvent(Trade event) throws Exception {
//这里做具体的消费逻辑
event.setId(UUID.randomUUID().toString());//简单生成下ID
System.out.println(event.getId());
}
}
public class Trade {
private String id;//ID
private String name;
private double price;//金额
private AtomicInteger count = new AtomicInteger(0);
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;
}
public AtomicInteger getCount() {
return count;
}
public void setCount(AtomicInteger count) {
this.count = count;
}
}
- 使用WorkerPool消息处理器:
public class Main2 {
public static void main(String[] args) throws InterruptedException {
int BUFFER_SIZE=1024;
int THREAD_NUMBERS=4;
EventFactory<Trade> eventFactory = new EventFactory<Trade>() {
public Trade newInstance() {
return new Trade();
}
};
RingBuffer<Trade> ringBuffer = RingBuffer.createSingleProducer(eventFactory, BUFFER_SIZE);
SequenceBarrier sequenceBarrier = ringBuffer.newBarrier();
ExecutorService executor = Executors.newFixedThreadPool(THREAD_NUMBERS);
WorkHandler<Trade> handler = new TradeHandler();
WorkerPool<Trade> workerPool = new WorkerPool<Trade>(ringBuffer, sequenceBarrier, new IgnoreExceptionHandler(), handler);
workerPool.start(executor);
//下面这个生产8个数据
for(int i=0;i<8;i++){
long seq=ringBuffer.next();
ringBuffer.get(seq).setPrice(Math.random()*9999);
ringBuffer.publish(seq);
}
Thread.sleep(1000);
workerPool.halt();
executor.shutdown();
}
}
public class TradeHandler implements EventHandler<Trade>, WorkHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
this.onEvent(event);
}
@Override
public void onEvent(Trade event) throws Exception {
//这里做具体的消费逻辑
event.setId(UUID.randomUUID().toString());//简单生成下ID
System.out.println(event.getId());
}
}
public class Trade {
private String id;//ID
private String name;
private double price;//金额
private AtomicInteger count = new AtomicInteger(0);
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;
}
public AtomicInteger getCount() {
return count;
}
public void setCount(AtomicInteger count) {
this.count = count;
}
}
- 在复杂场景下使用RingBuffer(希望P1生产的数据给C1、C2并行执行,最后C1、C2执行结束后C3执行):
public class Main {
public static void main(String[] args) throws InterruptedException {
long beginTime=System.currentTimeMillis();
int bufferSize=1024;
ExecutorService executor=Executors.newFixedThreadPool(8);
Disruptor<Trade> disruptor = new Disruptor<Trade>(new EventFactory<Trade>() {
@Override
public Trade newInstance() {
return new Trade();
}
}, bufferSize, executor, ProducerType.SINGLE, new BusySpinWaitStrategy());
//菱形操作
/**
//使用disruptor创建消费者组C1,C2
EventHandlerGroup<Trade> handlerGroup =
disruptor.handleEventsWith(new Handler1(), new Handler2());
//声明在C1,C2完事之后执行JMS消息发送操作 也就是流程走到C3
handlerGroup.then(new Handler3());
*/
//顺序操作
/**
disruptor.handleEventsWith(new Handler1()).
handleEventsWith(new Handler2()).
handleEventsWith(new Handler3());
*/
//六边形操作.
/**
Handler1 h1 = new Handler1();
Handler2 h2 = new Handler2();
Handler3 h3 = new Handler3();
Handler4 h4 = new Handler4();
Handler5 h5 = new Handler5();
disruptor.handleEventsWith(h1, h2);
disruptor.after(h1).handleEventsWith(h4);
disruptor.after(h2).handleEventsWith(h5);
disruptor.after(h4, h5).handleEventsWith(h3);
*/
disruptor.start();//启动
CountDownLatch latch=new CountDownLatch(1);
//生产者准备
executor.submit(new TradePublisher(latch, disruptor));
latch.await();//等待生产者完事.
disruptor.shutdown();
executor.shutdown();
System.out.println("总耗时:"+(System.currentTimeMillis()-beginTime));
}
}
public class Handler1 implements EventHandler<Trade>,WorkHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
this.onEvent(event);
}
@Override
public void onEvent(Trade event) throws Exception {
System.out.println("handler1: set name");
event.setName("h1");
Thread.sleep(1000);
}
}
public class Handler2 implements EventHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
System.out.println("handler2: set price");
event.setPrice(17.0);
Thread.sleep(1000);
}
}
public class Handler3 implements EventHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
System.out.println("handler3: name: " + event.getName() + " , price: " + event.getPrice() + "; instance: " + event.toString());
}
}
public class Handler4 implements EventHandler<Trade>,WorkHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
this.onEvent(event);
}
@Override
public void onEvent(Trade event) throws Exception {
System.out.println("handler4: get name : " + event.getName());
event.setName(event.getName() + "h4");
}
}
public class Handler5 implements EventHandler<Trade>,WorkHandler<Trade> {
@Override
public void onEvent(Trade event, long sequence, boolean endOfBatch) throws Exception {
this.onEvent(event);
}
@Override
public void onEvent(Trade event) throws Exception {
System.out.println("handler5: get price : " + event.getPrice());
event.setPrice(event.getPrice() + 3.0);
}
}
public class TradePublisher implements Runnable {
Disruptor<Trade> disruptor;
private CountDownLatch latch;
private static int LOOP=10;//模拟百万次交易的发生
public TradePublisher(CountDownLatch latch,Disruptor<Trade> disruptor) {
this.disruptor=disruptor;
this.latch=latch;
}
@Override
public void run() {
TradeEventTranslator tradeTransloator = new TradeEventTranslator();
for(int i=0;i<LOOP;i++){
disruptor.publishEvent(tradeTransloator);
}
latch.countDown();
}
}
class TradeEventTranslator implements EventTranslator<Trade>{
private Random random=new Random();
@Override
public void translateTo(Trade event, long sequence) {
this.generateTrade(event);
}
private Trade generateTrade(Trade trade){
trade.setPrice(random.nextDouble()*9999);
return trade;
}
}
- 多生产者和多消费者(复杂场景使用Disruptor,简单场景使用RingBuffer即可):
public class Main {
public static void main(String[] args) throws Exception {
//创建ringBuffer
RingBuffer<Order> ringBuffer =
RingBuffer.create(ProducerType.MULTI,
new EventFactory<Order>() {
@Override
public Order newInstance() {
return new Order();
}
},
1024 * 1024,
new YieldingWaitStrategy());
SequenceBarrier barriers = ringBuffer.newBarrier();
Consumer[] consumers = new Consumer[3];
for(int i = 0; i < consumers.length; i++){
consumers[i] = new Consumer("c" + i);
}
WorkerPool<Order> workerPool =
new WorkerPool<Order>(ringBuffer,
barriers,
new IntEventExceptionHandler(),
consumers);
ringBuffer.addGatingSequences(workerPool.getWorkerSequences());
workerPool.start(Executors.newFixedThreadPool(Runtime.getRuntime().availableProcessors()));
final CountDownLatch latch = new CountDownLatch(1);
for (int i = 0; i < 100; i++) {
final Producer p = new Producer(ringBuffer);
new Thread(new Runnable() {
@Override
public void run() {
try {
latch.await();
} catch (InterruptedException e) {
e.printStackTrace();
}
for(int j = 0; j < 100; j ++){
p.onData(UUID.randomUUID().toString());
}
}
}).start();
}
Thread.sleep(2000);
System.out.println("---------------开始生产-----------------");
latch.countDown();
Thread.sleep(5000);
System.out.println("总数:" + consumers[0].getCount() );
}
static class IntEventExceptionHandler implements ExceptionHandler {
public void handleEventException(Throwable ex, long sequence, Object event) {}
public void handleOnStartException(Throwable ex) {}
public void handleOnShutdownException(Throwable ex) {}
}
}
public class Order {
private String id;//ID
private String name;
private double price;//金额
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;
}
}
public class Producer {
private final RingBuffer<Order> ringBuffer;
public Producer(RingBuffer<Order> ringBuffer){
this.ringBuffer = ringBuffer;
}
/**
* onData用来发布事件,每调用一次就发布一次事件
* 它的参数会用过事件传递给消费者
*/
public void onData(String data){
//可以把ringBuffer看做一个事件队列,那么next就是得到下面一个事件槽
long sequence = ringBuffer.next();
try {
//用上面的索引取出一个空的事件用于填充(获取该序号对应的事件对象)
Order order = ringBuffer.get(sequence);
//获取要通过事件传递的业务数据
order.setId(data);
} finally {
//发布事件
//注意,最后的 ringBuffer.publish 方法必须包含在 finally 中以确保必须得到调用;如果某个请求的 sequence 未被提交,将会堵塞后续的发布操作或者其它的 producer。
ringBuffer.publish(sequence);
}
}
}
public class Consumer implements WorkHandler<Order>{
private String consumerId;
private static AtomicInteger count = new AtomicInteger(0);
public Consumer(String consumerId){
this.consumerId = consumerId;
}
@Override
public void onEvent(Order order) throws Exception {
System.out.println("当前消费者: " + this.consumerId + ",消费信息:" + order.getId());
count.incrementAndGet();
}
public int getCount(){
return count.get();
}
}