java web解决瞬间高并发的策略

1、任何的高并发，请求总是会有一个顺序的

2、java的队列的数据结构是先进先出的取值顺序

3、BlockingQueue类(线程安全)（使用方法可以百度）

一般使用LinkedBlockingQueue

利用以上几点，我们可以把高并发时候的请求放入一个队列，队列的大小可以自己定义，比如队列容量为1000个数据，那么可以利用过滤器或者拦截器把当前的请求放入队列，如果队列的容量满了，其余的请求可以丢掉或者作出相应回复

具体实施：

利用生产者、消费者模型：

将队列的请求一一处理完。

上代码：

/**
 * @author fuguangli
 * @description 前沿消费者类
 * @Create date:    2017/3/7
 * @using   EXAMPLE
 */
public class Customer implements Runnable{


    /**
     *         抛出异常    特殊值        阻塞         超时
     插入    add(e)    offer(e)    put(e)    offer(e, time, unit)
     移除    remove()    poll()    take()    poll(time, unit)
     检查    element()    peek()    不可用    不可用

     */
    private BlockingQueue blockingQueue;
    private AtomicInteger count = new AtomicInteger();
    public Customer(BlockingQueue blockingQueue) {
        this.blockingQueue = blockingQueue;
    }

    /**
     * When an object implementing interface <code>Runnable</code> is used
     * to create a thread, starting the thread causes the object's
     * <code>run</code> method to be called in that separately executing
     * thread.
     * <p/>
     * The general contract of the method <code>run</code> is that it may
     * take any action whatsoever.
     *
     * @see Thread#run()
     */
    @Override
    public void run() {
        System.out.println("消费者线程启动...");
        LockFlag.setCustomerRunningFlag(true);
        try {
            while (LockFlag.getProducerRunningFlag()){
                System.out.println(Thread.currentThread().getId()+"I'm Customer.Queue current size="+blockingQueue.size());
                String data = (String) blockingQueue.poll(10, TimeUnit.SECONDS);
                if(data!=null){
                    System.out.println(Thread.currentThread().getId()+"*************正在消费数据 data="+data);
                }else{
                    //表示超过取值时间，视为生产者不再生产数据
                    System.out.println(Thread.currentThread().getId()+"队列为空无数据，请检查生产者是否阻塞");
                }
                Thread.sleep(50);
            }
            System.err.println("消费者程序执行完毕");
        } catch (InterruptedException e) {
            e.printStackTrace();
            System.err.println("消费者程序退出");
            LockFlag.setCustomerRunningFlag(false);//异常退出线程
            Thread.currentThread().interrupt();
        }
    }
}

package com.qysxy.framework.queue;

import java.util.concurrent.BlockingQueue;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicInteger;

/**
 * @author fuguangli
 * @description 队列生产者类
 * @Create date:    2017/3/7
 * @using       EXAMPLE
 */
public class Producer implements Runnable{


    /**
     *         抛出异常    特殊值        阻塞         超时
     插入    add(e)    offer(e)    put(e)    offer(e, time, unit)
     移除    remove()    poll()    take()    poll(time, unit)
     检查    element()    peek()    不可用    不可用

     */
    private BlockingQueue blockingQueue;
    private AtomicInteger count = new AtomicInteger();
    public Producer(BlockingQueue blockingQueue) {
        this.blockingQueue = blockingQueue;
    }

    /**
     * When an object implementing interface <code>Runnable</code> is used
     * to create a thread, starting the thread causes the object's
     * <code>run</code> method to be called in that separately executing
     * thread.
     * <p/>
     * The general contract of the method <code>run</code> is that it may
     * take any action whatsoever.
     *
     * @see Thread#run()
     */
    @Override
    public void run() {
        System.out.println("生产者线程启动...");
        LockFlag.setProducerRunningFlag(true);
        try {
            while (LockFlag.getProducerRunningFlag()){
                String data = "data:"+count.incrementAndGet();
                if(blockingQueue.offer(data,10, TimeUnit.SECONDS)){
                    //返回true表示生产数据正确
                    System.out.println("^^^^^^^^^^^^^^正在生产数据 data="+data);
                }else {
                    //表示阻塞时间内还没有生产者生产数据
                    System.out.println("生产者异常，无法生产数据");
                }
                Thread.sleep(50);

            }
        } catch (InterruptedException e) {
            e.printStackTrace();
            System.err.println("生产者程序退出");
            LockFlag.setProducerRunningFlag(false);//异常退出线程
            Thread.currentThread().interrupt();
        }
    }
}

package com.qysxy.framework.queue;

/**
 * @author fuguangli
 * @description 前沿生产者消费者模型的锁类
 * @Create date:    2017/3/7
 */
public class LockFlag {
    /**
     * 生产者互斥锁
     */
    private static Boolean producerRunningFlag = false;
    /**
     * 消费者互斥锁
     */
    private static Boolean customerRunningFlag = false;

    public static Boolean getProducerRunningFlag() {
        return producerRunningFlag;
    }

    public static void setProducerRunningFlag(Boolean producerRunningFlag) {
        LockFlag.producerRunningFlag = producerRunningFlag;
    }

    public static Boolean getCustomerRunningFlag() {
        return customerRunningFlag;
    }

    public static void setCustomerRunningFlag(Boolean customerRunningFlag) {
        LockFlag.customerRunningFlag = customerRunningFlag;
    }
}

package com.qysxy.framework.queue;

import javax.servlet.http.HttpServletRequest;
import javax.servlet.http.HttpServletResponse;
import java.util.Queue;
import java.util.concurrent.*;

/**
 * @author fuguangli
 * @description 前沿队列实用类，用于大量并发用户
 * @Create date:    2017/3/7
 */
public class BlockingQueueHelper {


    private static final Integer maxQueueSize = 1000;
    private static BlockingQueue blockingQueue = new LinkedBlockingQueue(maxQueueSize);
    private static ExecutorService threadPool = Executors.newCachedThreadPool();


    public static BlockingQueue getBlockingQueue() {
        if (blockingQueue == null) {
            blockingQueue = new LinkedBlockingQueue(maxQueueSize);
        }
        return blockingQueue;
    }

    /**
     * @param o 队列处理对象（包含request,response,data）
     */
    public static void requestQueue(Object o) {
        //检测当前的队列大小
        if (blockingQueue != null && blockingQueue.size() < maxQueueSize) {
            //可以正常进入队列
            if (blockingQueue.offer(o)) {
                //添加成功，检测数据处理线程是否正常
                if (LockFlag.getCustomerRunningFlag()) {
                    //说明处理线程类正常运行
                } else {
                    //说明处理线程类停止，此时，应重新启动线程进行数据处理
                    LockFlag.setCustomerRunningFlag(true);

                    //example:run
                    Customer customer = new Customer(blockingQueue);
                    threadPool.execute(customer);

                }

            } else {
                //进入队列失败，做出相应的处理，或者尝试重新进入队列

            }
        } else {
            //队列不正常，或队列大小已达上限，做出相应处理

        }

    }
}

好了，这时候，利用过滤器或者拦截器将每个请求封装成队列元素进行处理就行。

当然了，对于多应用服务器的部署架构来说，数据库也需要加锁，数据库隔离级别下篇再说。