• 分布式全局不重复ID生成算法


    在分布式系统中经常会使用到生成全局唯一不重复ID的情况。本篇博客介绍生成的一些方法。 

    常见的一些方式: 

    1、通过DB做全局自增操作 
    优点:简单、高效 
    缺点:大并发、分布式情况下性能比较低 

    有些同学可能会说分库、分表的策略去降低DB的瓶颈,单要做到全局不重复需要提前按照一定的区域进行划分。例如:1~10000、10001~20000 等等。但这个灵活度比较低。 

    针对一些并发比较低的情况也可以使用类似这种方式。但大并发时不建议使用,DB很容易成为瓶颈。 

    2、获取当前时间纳秒或毫秒数 
    这种方式需要考虑的是在分布式集群中如果保证唯一性。 

    3、类似UUID的生成方式 
    生成的串比较大 

    //------------------------------------------------------------ 
    综合上述情况我们需要一种在高并发、分布式系统中提供高效生成不重复唯一的一个ID,但要求生成的结果要小 
    方法1: 
    private static long INFOID_FLAG = 1260000000000L; 
    protected static int SERVER_ID = 1; 

    public synchronized long nextId() throws Exception { 
        if(SERVER_ID <= 0) 
            throw new Exception("server id is error,please check config file!"); 
        long infoid = System.currentTimeMillis() - INFOID_FLAG; 
        infoid=(infoid<<7)| SERVER_ID; 
        Thread.sleep(1); 
        return infoid; 


    说明: 
    SERVER_ID为不同的服务器使用的不同server ID,如果不同的机器使用相同的server ID有可能会生成重复的全局ID 

    简单的应用在一定的并发情况下使用这种方式已经足够了,简单、高效。但是每秒生成的ID是有限的,因为Thread.sleep(1)会无形中带来一些时间的消耗。 

    方法2: 


    /** 
    * 64位ID (42(毫秒)+5(机器ID)+5(业务编码)+12(重复累加)) 
    * @author Polim 
    */ 
    public class IdWorker { 
    private final static long twepoch = 1288834974657L; 
    // 机器标识位数 
    private final static long workerIdBits = 5L; 
    // 数据中心标识位数 
    private final static long datacenterIdBits = 5L; 
    // 机器ID最大值 
    private final static long maxWorkerId = -1L ^ (-1L << workerIdBits); 
    // 数据中心ID最大值 
    private final static long maxDatacenterId = -1L ^ (-1L << datacenterIdBits); 
    // 毫秒内自增位 
    private final static long sequenceBits = 12L; 
    // 机器ID偏左移12位 
    private final static long workerIdShift = sequenceBits; 
    // 数据中心ID左移17位 
    private final static long datacenterIdShift = sequenceBits + workerIdBits; 
    // 时间毫秒左移22位 
    private final static long timestampLeftShift = sequenceBits + workerIdBits + datacenterIdBits; 

    private final static long sequenceMask = -1L ^ (-1L << sequenceBits); 

    private static long lastTimestamp = -1L; 

    private long sequence = 0L; 
    private final long workerId; 
    private final long datacenterId; 

    public IdWorker(long workerId, long datacenterId) { 
    if (workerId > maxWorkerId || workerId < 0) { 
    throw new IllegalArgumentException("worker Id can't be greater than %d or less than 0"); 

    if (datacenterId > maxDatacenterId || datacenterId < 0) { 
    throw new IllegalArgumentException("datacenter Id can't be greater than %d or less than 0"); 

    this.workerId = workerId; 
    this.datacenterId = datacenterId; 


    public synchronized long nextId() { 
    long timestamp = timeGen(); 
    if (timestamp < lastTimestamp) { 
    try { 
    throw new Exception("Clock moved backwards.  Refusing to generate id for "+ (lastTimestamp - timestamp) + " milliseconds"); 
    } catch (Exception e) { 
    e.printStackTrace(); 



    if (lastTimestamp == timestamp) { 
    // 当前毫秒内,则+1 
    sequence = (sequence + 1) & sequenceMask; 
    if (sequence == 0) { 
    // 当前毫秒内计数满了,则等待下一秒 
    timestamp = tilNextMillis(lastTimestamp); 

    } else { 
    sequence = 0; 

    lastTimestamp = timestamp; 
    // ID偏移组合生成最终的ID,并返回ID 
    long nextId = ((timestamp - twepoch) << timestampLeftShift) 
    | (datacenterId << datacenterIdShift) 
    | (workerId << workerIdShift) | sequence; 

    return nextId; 


    private long tilNextMillis(final long lastTimestamp) { 
    long timestamp = this.timeGen(); 
    while (timestamp <= lastTimestamp) { 
    timestamp = this.timeGen(); 

    return timestamp; 


    private long timeGen() { 
    return System.currentTimeMillis(); 



    这种方式是一种比较高效的方式。也是twitter使用的一种方式。 

    测试类:---------------------------------------------------------- 
    import java.util.concurrent.BrokenBarrierException; 
    import java.util.concurrent.CountDownLatch; 
    import java.util.concurrent.CyclicBarrier; 
    import java.util.concurrent.TimeUnit; 

    public class IdWorkerTest { 
        public static void main(String []args){ 
            IdWorkerTest test = new IdWorkerTest(); 
            test.test2(); 
        } 

        public void test2(){ 
            final IdWorker w = new IdWorker(1,2); 
            final CyclicBarrier cdl = new CyclicBarrier(100); 

            for(int i = 0; i < 100; i++){ 
                new Thread(new Runnable() { 
                    @Override 
                    public void run() { 
                    try { 
                        cdl.await(); 
                    } catch (InterruptedException e) { 
                        e.printStackTrace(); 
                    } catch (BrokenBarrierException e) { 
                        e.printStackTrace(); 
                    } 
                    System.out.println(w.nextId());} 
                 }).start(); 
            } 
            try { 
                TimeUnit.SECONDS.sleep(5); 
            } catch (InterruptedException e) { 
               e.printStackTrace(); 
            } 

        } 
    }
  • 相关阅读:
    Python GUI编程(Tkinter)19、Frame控件
    Python GUI编程(Tkinter)18、Combobox下拉控件
    D
    C
    B
    A
    wordpress调用服务器本地的头像
    杂七杂八的问题处理03--jenkins发邮件提示Error sending to the following VALID addresses
    杂七杂八的问题处理02--allure报告显示loading问题
    vue一次下载多个文件
  • 原文地址:https://www.cnblogs.com/firstdream/p/5222520.html
Copyright © 2020-2023  润新知