理解分布式id生成算法SnowFlake

分布式id生成算法的有很多种,Twitter的SnowFlake就是其中经典的一种。

概述

SnowFlake算法生成id的结果是一个64bit大小的整数,它的结构如下图:

理解分布式id生成算法SnowFlake

  • 1位,不用。二进制中最高位为1的都是负数,但是我们生成的id一般都使用整数,所以这个最高位固定是0
  • 41位,用来记录时间戳(毫秒)。

    • 41位可以表示$2^{41}-1$个数字,
    • 如果只用来表示正整数(计算机中正数包含0),可以表示的数值范围是:0 至 $2^{41}-1$,减1是因为可表示的数值范围是从0开始算的,而不是1。
    • 也就是说41位可以表示$2^{41}-1$个毫秒的值,转化成单位年则是$(2^{41}-1) / (1000 * 60 * 60 * 24 * 365) = 69$年
  • 10位,用来记录工作机器id。

    • 可以部署在$2^{10} = 1024$个节点,包括5位datacenterId5位workerId
    • 5位(bit)可以表示的最大正整数是$2^{5}-1 = 31$,即可以用0、1、2、3、....31这32个数字,来表示不同的datecenterId或workerId
  • 12位,序列号,用来记录同毫秒内产生的不同id。

    • 12位(bit)可以表示的最大正整数是$2^{12}-1 = 4095$,即可以用0、1、2、3、....4094这4095个数字,来表示同一机器同一时间截(毫秒)内产生的4095个ID序号

由于在Java中64bit的整数是long类型,所以在Java中SnowFlake算法生成的id就是long来存储的。

SnowFlake可以保证:

  • 所有生成的id按时间趋势递增
  • 整个分布式系统内不会产生重复id(因为有datacenterId和workerId来做区分)

Talk is cheap, show you the code

以下是Twitter官方原版的,用Scala写的,(我也不懂Scala,当成Java看即可):

/** Copyright 2010-2012 Twitter, Inc.*/
package com.twitter.service.snowflake

import com.twitter.ostrich.stats.Stats
import com.twitter.service.snowflake.gen._
import java.util.Random
import com.twitter.logging.Logger

/**
 * An object that generates IDs.
 * This is broken into a separate class in case
 * we ever want to support multiple worker threads
 * per process
 */
class IdWorker(
    val workerId: Long, 
    val datacenterId: Long, 
    private val reporter: Reporter, 
    var sequence: Long = 0L) extends Snowflake.Iface {
    
  private[this] def genCounter(agent: String) = {
    Stats.incr("ids_generated")
    Stats.incr("ids_generated_%s".format(agent))
  }
  private[this] val exceptionCounter = Stats.getCounter("exceptions")
  private[this] val log = Logger.get
  private[this] val rand = new Random

  val twepoch = 1288834974657L

  private[this] val workerIdBits = 5L
  private[this] val datacenterIdBits = 5L
  private[this] val maxWorkerId = -1L ^ (-1L << workerIdBits)
  private[this] val maxDatacenterId = -1L ^ (-1L << datacenterIdBits)
  private[this] val sequenceBits = 12L

  private[this] val workerIdShift = sequenceBits
  private[this] val datacenterIdShift = sequenceBits + workerIdBits
  private[this] val timestampLeftShift = sequenceBits + workerIdBits + datacenterIdBits
  private[this] val sequenceMask = -1L ^ (-1L << sequenceBits)

  private[this] var lastTimestamp = -1L

  // sanity check for workerId
  if (workerId > maxWorkerId || workerId < 0) {
    exceptionCounter.incr(1)
    throw new IllegalArgumentException("worker Id can't be greater than %d or less than 0".format(maxWorkerId))
  }

  if (datacenterId > maxDatacenterId || datacenterId < 0) {
    exceptionCounter.incr(1)
    throw new IllegalArgumentException("datacenter Id can't be greater than %d or less than 0".format(maxDatacenterId))
  }

  log.info("worker starting. timestamp left shift %d, datacenter id bits %d, worker id bits %d, sequence bits %d, workerid %d",
    timestampLeftShift, datacenterIdBits, workerIdBits, sequenceBits, workerId)

  def get_id(useragent: String): Long = {
    if (!validUseragent(useragent)) {
      exceptionCounter.incr(1)
      throw new InvalidUserAgentError
    }

    val id = nextId()
    genCounter(useragent)

    reporter.report(new AuditLogEntry(id, useragent, rand.nextLong))
    id
  }

  def get_worker_id(): Long = workerId
  def get_datacenter_id(): Long = datacenterId
  def get_timestamp() = System.currentTimeMillis

  protected[snowflake] def nextId(): Long = synchronized {
    var timestamp = timeGen()

    if (timestamp < lastTimestamp) {
      exceptionCounter.incr(1)
      log.error("clock is moving backwards.  Rejecting requests until %d.", lastTimestamp);
      throw new InvalidSystemClock("Clock moved backwards.  Refusing to generate id for %d milliseconds".format(
        lastTimestamp - timestamp))
    }

    if (lastTimestamp == timestamp) {
      sequence = (sequence + 1) & sequenceMask
      if (sequence == 0) {
        timestamp = tilNextMillis(lastTimestamp)
      }
    } else {
      sequence = 0
    }

    lastTimestamp = timestamp
    ((timestamp - twepoch) << timestampLeftShift) |
      (datacenterId << datacenterIdShift) |
      (workerId << workerIdShift) | 
      sequence
  }

  protected def tilNextMillis(lastTimestamp: Long): Long = {
    var timestamp = timeGen()
    while (timestamp <= lastTimestamp) {
      timestamp = timeGen()
    }
    timestamp
  }

  protected def timeGen(): Long = System.currentTimeMillis()

  val AgentParser = """([a-zA-Z][a-zA-Z\-0-9]*)""".r

  def validUseragent(useragent: String): Boolean = useragent match {
    case AgentParser(_) => true
    case _ => false
  }
}

Scala是一门可以编译成字节码的语言,简单理解是在Java语法基础上加上了很多语法糖,例如不用每条语句后写分号,可以使用动态类型等等。抱着试一试的心态,我把Scala版的代码“翻译”成Java版本的,对scala代码改动的地方如下:

/** Copyright 2010-2012 Twitter, Inc.*/
package com.twitter.service.snowflake

import com.twitter.ostrich.stats.Stats 
import com.twitter.service.snowflake.gen._
import java.util.Random
import com.twitter.logging.Logger

/**
 * An object that generates IDs.
 * This is broken into a separate class in case
 * we ever want to support multiple worker threads
 * per process
 */
class IdWorker(                                        // |
    val workerId: Long,                                // |
    val datacenterId: Long,                            // |<--这部分改成Java的构造函数形式
    private val reporter: Reporter,//日志相关,删       // |
    var sequence: Long = 0L)                           // |
       extends Snowflake.Iface { //接口找不到,删       // |     
    
  private[this] def genCounter(agent: String) = {                     // |
    Stats.incr("ids_generated")                                       // |
    Stats.incr("ids_generated_%s".format(agent))                      // |<--错误、日志处理相关,删
  }                                                                   // | 
  private[this] val exceptionCounter = Stats.getCounter("exceptions") // |
  private[this] val log = Logger.get                                  // |
  private[this] val rand = new Random                                 // | 

  val twepoch = 1288834974657L

  private[this] val workerIdBits = 5L
  private[this] val datacenterIdBits = 5L
  private[this] val maxWorkerId = -1L ^ (-1L << workerIdBits)
  private[this] val maxDatacenterId = -1L ^ (-1L << datacenterIdBits)
  private[this] val sequenceBits = 12L

  private[this] val workerIdShift = sequenceBits
  private[this] val datacenterIdShift = sequenceBits + workerIdBits
  private[this] val timestampLeftShift = sequenceBits + workerIdBits + datacenterIdBits
  private[this] val sequenceMask = -1L ^ (-1L << sequenceBits)

  private[this] var lastTimestamp = -1L

  //----------------------------------------------------------------------------------------------------------------------------//
  // sanity check for workerId                                                                                                  //
  if (workerId > maxWorkerId || workerId < 0) {                                                                                 //
    exceptionCounter.incr(1) //<--错误处理相关,删                                                                               //
    throw new IllegalArgumentException("worker Id can't be greater than %d or less than 0".format(maxWorkerId))                 //这
    // |-->改成:throw new IllegalArgumentException                                                                              //部
    //            (String.format("worker Id can't be greater than %d or less than 0",maxWorkerId))                              //分
  }                                                                                                                             //放
                                                                                                                                //到
  if (datacenterId > maxDatacenterId || datacenterId < 0) {                                                                     //构
    exceptionCounter.incr(1) //<--错误处理相关,删                                                                               //造
    throw new IllegalArgumentException("datacenter Id can't be greater than %d or less than 0".format(maxDatacenterId))         //函
    // |-->改成:throw new IllegalArgumentException                                                                             //数
    //             (String.format("datacenter Id can't be greater than %d or less than 0",maxDatacenterId))                     //中
  }                                                                                                                             //
                                                                                                                                //
  log.info("worker starting. timestamp left shift %d, datacenter id bits %d, worker id bits %d, sequence bits %d, workerid %d", //  
    timestampLeftShift, datacenterIdBits, workerIdBits, sequenceBits, workerId)                                                 //   
  // |-->改成:System.out.printf("worker...%d...",timestampLeftShift,...);                                                      //
  //----------------------------------------------------------------------------------------------------------------------------//

  //-------------------------------------------------------------------//  
  //这个函数删除错误处理相关的代码后,剩下一行代码:val id = nextId()      //
  //所以我们直接调用nextId()函数可以了,所以在“翻译”时可以删除这个函数      //
  def get_id(useragent: String): Long = {                              // 
    if (!validUseragent(useragent)) {                                  //
      exceptionCounter.incr(1)                                         //
      throw new InvalidUserAgentError                                  //删
    }                                                                  //除
                                                                       // 
    val id = nextId()                                                  // 
    genCounter(useragent)                                              //
                                                                       //
    reporter.report(new AuditLogEntry(id, useragent, rand.nextLong))   //
    id                                                                 //
  }                                                                    // 
  //-------------------------------------------------------------------//

  def get_worker_id(): Long = workerId           // |
  def get_datacenter_id(): Long = datacenterId   // |<--改成Java函数
  def get_timestamp() = System.currentTimeMillis // |

  protected[snowflake] def nextId(): Long = synchronized { // 改成Java函数
    var timestamp = timeGen()

    if (timestamp < lastTimestamp) {
      exceptionCounter.incr(1) // 错误处理相关,删
      log.error("clock is moving backwards.  Rejecting requests until %d.", lastTimestamp); // 改成System.err.printf(...)
      throw new InvalidSystemClock("Clock moved backwards.  Refusing to generate id for %d milliseconds".format(
        lastTimestamp - timestamp)) // 改成RumTimeException
    }

    if (lastTimestamp == timestamp) {
      sequence = (sequence + 1) & sequenceMask
      if (sequence == 0) {
        timestamp = tilNextMillis(lastTimestamp)
      }
    } else {
      sequence = 0
    }

    lastTimestamp = timestamp
    ((timestamp - twepoch) << timestampLeftShift) | // |<--加上关键字return
      (datacenterId << datacenterIdShift) |         // |
      (workerId << workerIdShift) |                 // |
      sequence                                      // |
  }

  protected def tilNextMillis(lastTimestamp: Long): Long = { // 改成Java函数
    var timestamp = timeGen()
    while (timestamp <= lastTimestamp) {
      timestamp = timeGen()
    }
    timestamp // 加上关键字return
  }

  protected def timeGen(): Long = System.currentTimeMillis() // 改成Java函数

  val AgentParser = """([a-zA-Z][a-zA-Z\-0-9]*)""".r                  // |
                                                                      // | 
  def validUseragent(useragent: String): Boolean = useragent match {  // |<--日志相关,删
    case AgentParser(_) => true                                       // |
    case _ => false                                                   // |   
  }                                                                   // | 
}

改出来的Java版:

public class IdWorker{

    private long workerId;
    private long datacenterId;
    private long sequence;

    public IdWorker(long workerId, long datacenterId, long sequence){
        // sanity check for workerId
        if (workerId > maxWorkerId || workerId < 0) {
            throw new IllegalArgumentException(String.format("worker Id can't be greater than %d or less than 0",maxWorkerId));
        }
        if (datacenterId > maxDatacenterId || datacenterId < 0) {
            throw new IllegalArgumentException(String.format("datacenter Id can't be greater than %d or less than 0",maxDatacenterId));
        }
        System.out.printf("worker starting. timestamp left shift %d, datacenter id bits %d, worker id bits %d, sequence bits %d, workerid %d",
                timestampLeftShift, datacenterIdBits, workerIdBits, sequenceBits, workerId);

        this.workerId = workerId;
        this.datacenterId = datacenterId;
        this.sequence = sequence;
    }

    private long twepoch = 1288834974657L;

    private long workerIdBits = 5L;
    private long datacenterIdBits = 5L;
    private long maxWorkerId = -1L ^ (-1L << workerIdBits);
    private long maxDatacenterId = -1L ^ (-1L << datacenterIdBits);
    private long sequenceBits = 12L;

    private long workerIdShift = sequenceBits;
    private long datacenterIdShift = sequenceBits + workerIdBits;
    private long timestampLeftShift = sequenceBits + workerIdBits + datacenterIdBits;
    private long sequenceMask = -1L ^ (-1L << sequenceBits);

    private long lastTimestamp = -1L;

    public long getWorkerId(){
        return workerId;
    }

    public long getDatacenterId(){
        return datacenterId;
    }

    public long getTimestamp(){
        return System.currentTimeMillis();
    }

    public synchronized long nextId() {
        long timestamp = timeGen();

        if (timestamp < lastTimestamp) {
            System.err.printf("clock is moving backwards.  Rejecting requests until %d.", lastTimestamp);
            throw new RuntimeException(String.format("Clock moved backwards.  Refusing to generate id for %d milliseconds",
                    lastTimestamp - timestamp));
        }

        if (lastTimestamp == timestamp) {
            sequence = (sequence + 1) & sequenceMask;
            if (sequence == 0) {
                timestamp = tilNextMillis(lastTimestamp);
            }
        } else {
            sequence = 0;
        }

        lastTimestamp = timestamp;
        return ((timestamp - twepoch) << timestampLeftShift) |
                (datacenterId << datacenterIdShift) |
                (workerId << workerIdShift) |
                sequence;
    }

    private long tilNextMillis(long lastTimestamp) {
        long timestamp = timeGen();
        while (timestamp <= lastTimestamp) {
            timestamp = timeGen();
        }
        return timestamp;
    }

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

    //---------------测试---------------
    public static void main(String[] args) {
        IdWorker worker = new IdWorker(1,1,1);
        for (int i = 0; i < 30; i++) {
            System.out.println(worker.nextId());
        }
    }

}

代码理解

上面的代码中,有部分位运算的代码,如:

sequence = (sequence + 1) & sequenceMask;

private long maxWorkerId = -1L ^ (-1L << workerIdBits);

return ((timestamp - twepoch) << timestampLeftShift) |
        (datacenterId << datacenterIdShift) |
        (workerId << workerIdShift) |
        sequence;

为了能更好理解,我对相关知识研究了一下。

负数的二进制表示

在计算机中,负数的二进制是用补码来表示的。
假设我是用Java中的int类型来存储数字的,
int类型的大小是32个二进制位(bit),即4个字节(byte)。(1 byte = 8 bit)
那么十进制数字3在二进制中的表示应该是这样的:

00000000 00000000 00000000 00000011
// 3的二进制表示,就是原码

那数字-3在二进制中应该如何表示?
我们可以反过来想想,因为-3+3=0,
在二进制运算中把-3的二进制看成未知数x来求解
求解算式的二进制表示如下:

00000000 00000000 00000000 00000011 //3,原码
+  xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx //-3,补码
-----------------------------------------------
   00000000 00000000 00000000 00000000

反推x的值,3的二进制加上什么值才使结果变成00000000 00000000 00000000 00000000?:

00000000 00000000 00000000 00000011 //3,原码                         
+  11111111 11111111 11111111 11111101 //-3,补码
-----------------------------------------------
 1 00000000 00000000 00000000 00000000

反推的思路是3的二进制数从最低位开始逐位加1,使溢出的1不断向高位溢出,直到溢出到第33位。然后由于int类型最多只能保存32个二进制位,所以最高位的1溢出了,剩下的32位就成了(十进制的)0。

补码的意义就是可以拿补码和原码(3的二进制)相加,最终加出一个“溢出的0”

以上是理解的过程,实际中记住公式就很容易算出来:

  • 补码 = 反码 + 1
  • 补码 = (原码 - 1)再取反码

因此-1的二进制应该这样算:

00000000 00000000 00000000 00000001 //原码:1的二进制
11111111 11111111 11111111 11111110 //取反码:1的二进制的反码
11111111 11111111 11111111 11111111 //加1:-1的二进制表示(补码)

用位运算计算n个bit能表示的最大数值

比如这样一行代码:

private long workerIdBits = 5L;
    private long maxWorkerId = -1L ^ (-1L << workerIdBits);

上面代码换成这样看方便一点:
long maxWorkerId = -1L ^ (-1L << 5L)

咋一看真的看不准哪个部分先计算,于是查了一下Java运算符的优先级表:
理解分布式id生成算法SnowFlake

所以上面那行代码中,运行顺序是:

  • -1 左移 5,得结果a
  • -1 异或 a

long maxWorkerId = -1L ^ (-1L << 5L)的二进制运算过程如下:

-1 左移 5,得结果a :

11111111 11111111 11111111 11111111 //-1的二进制表示(补码)
  11111 11111111 11111111 11111111 11100000 //高位溢出的不要,低位补0
        11111111 11111111 11111111 11100000 //结果a

-1 异或 a :

11111111 11111111 11111111 11111111 //-1的二进制表示(补码)
    ^   11111111 11111111 11111111 11100000 //两个操作数的位中,相同则为0,不同则为1
---------------------------------------------------------------------------
        00000000 00000000 00000000 00011111 //最终结果31

最终结果是31,二进制00000000 00000000 00000000 00011111转十进制可以这么算:
$$ 2^4 + 2^3 + 2^2 + 2^1 + 2^0 = 16 + 8 + 4 + 2 + 1 =31 $$

那既然现在知道算出来long maxWorkerId = -1L ^ (-1L << 5L)中的maxWorkerId = 31,有什么含义?为什么要用左移5来算?如果你看过概述部分,请找到这段内容看看:

5位(bit)可以表示的最大正整数是$2^{5}-1 = 31$,即可以用0、1、2、3、....31这32个数字,来表示不同的datecenterId或workerId

-1L ^ (-1L << 5L)结果是31,$2^{5}-1$的结果也是31,所以在代码中,-1L ^ (-1L << 5L)的写法是利用位运算计算出5位能表示的最大正整数是多少

用mask防止溢出

有一段有趣的代码:

sequence = (sequence + 1) & sequenceMask;

分别用不同的值测试一下,你就知道它怎么有趣了:

long seqMask = -1L ^ (-1L << 12L); //计算12位能耐存储的最大正整数,相当于:2^12-1 = 4095
        System.out.println("seqMask: "+seqMask);
        System.out.println(1L & seqMask);
        System.out.println(2L & seqMask);
        System.out.println(3L & seqMask);
        System.out.println(4L & seqMask);
        System.out.println(4095L & seqMask);
        System.out.println(4096L & seqMask);
        System.out.println(4097L & seqMask);
        System.out.println(4098L & seqMask);

        
        /**
        seqMask: 4095
        1
        2
        3
        4
        4095
        0
        1
        2
        */

这段代码通过位与运算保证计算的结果范围始终是 0-4095 !

用位运算汇总结果

还有另外一段诡异的代码:

return ((timestamp - twepoch) << timestampLeftShift) |
        (datacenterId << datacenterIdShift) |
        (workerId << workerIdShift) |
        sequence;

为了弄清楚这段代码,

首先 需要计算一下相关的值:

private long twepoch = 1288834974657L; //起始时间戳,用于用当前时间戳减去这个时间戳,算出偏移量

    private long workerIdBits = 5L; //workerId占用的位数:5
    private long datacenterIdBits = 5L; //datacenterId占用的位数:5
    private long maxWorkerId = -1L ^ (-1L << workerIdBits);  // workerId可以使用的最大数值:31
    private long maxDatacenterId = -1L ^ (-1L << datacenterIdBits); // datacenterId可以使用的最大数值:31
    private long sequenceBits = 12L;//序列号占用的位数:12

    private long workerIdShift = sequenceBits; // 12
    private long datacenterIdShift = sequenceBits + workerIdBits; // 12+5 = 17
    private long timestampLeftShift = sequenceBits + workerIdBits + datacenterIdBits; // 12+5+5 = 22
    private long sequenceMask = -1L ^ (-1L << sequenceBits);//4095

    private long lastTimestamp = -1L;

其次 写个测试,把参数都写死,并运行打印信息,方便后面来核对计算结果:

//---------------测试---------------
    public static void main(String[] args) {
        long timestamp = 1505914988849L;
        long twepoch = 1288834974657L;
        long datacenterId = 17L;
        long workerId = 25L;
        long sequence = 0L;

        System.out.printf("\ntimestamp: %d \n",timestamp);
        System.out.printf("twepoch: %d \n",twepoch);
        System.out.printf("datacenterId: %d \n",datacenterId);
        System.out.printf("workerId: %d \n",workerId);
        System.out.printf("sequence: %d \n",sequence);
        System.out.println();
        System.out.printf("(timestamp - twepoch): %d \n",(timestamp - twepoch));
        System.out.printf("((timestamp - twepoch) << 22L): %d \n",((timestamp - twepoch) << 22L));
        System.out.printf("(datacenterId << 17L): %d \n" ,(datacenterId << 17L));
        System.out.printf("(workerId << 12L): %d \n",(workerId << 12L));
        System.out.printf("sequence: %d \n",sequence);

        long result = ((timestamp - twepoch) << 22L) |
                (datacenterId << 17L) |
                (workerId << 12L) |
                sequence;
        System.out.println(result);

    }

    /** 打印信息:
        timestamp: 1505914988849 
        twepoch: 1288834974657 
        datacenterId: 17 
        workerId: 25 
        sequence: 0 
        
        (timestamp - twepoch): 217080014192 
        ((timestamp - twepoch) << 22L): 910499571845562368 
        (datacenterId << 17L): 2228224 
        (workerId << 12L): 102400 
        sequence: 0 
        910499571847892992
    */

代入位移的值得之后,就是这样:

return ((timestamp - 1288834974657) << 22) |
        (datacenterId << 17) |
        (workerId << 12) |
        sequence;

对于尚未知道的值,我们可以先看看概述 中对SnowFlake结构的解释,再代入在合法范围的值(windows系统可以用计算器方便计算这些值的二进制),来了解计算的过程。
当然,由于我的测试代码已经把这些值写死了,那直接用这些值来手工验证计算结果即可:

long timestamp = 1505914988849L;
        long twepoch = 1288834974657L;
        long datacenterId = 17L;
        long workerId = 25L;
        long sequence = 0L;
设:timestamp  = 1505914988849,twepoch = 1288834974657
1505914988849 - 1288834974657 = 217080014192 (timestamp相对于起始时间的毫秒偏移量),其(a)二进制左移22位计算过程如下:                                

                        |<--这里开始左右22位                            ‭
00000000 00000000 000000|00 00110010 10001010 11111010 00100101 01110000 // a = 217080014192
00001100 10100010 10111110 10001001 01011100 00|000000 00000000 00000000 // a左移22位后的值(la)
                                               |<--这里后面的位补0

设:datacenterId  = 17,其(b)二进制左移17位计算过程如下:

                   |<--这里开始左移17位    
00000000 00000000 0|0000000 ‭00000000 00000000 00000000 00000000 00010001 // b = 17
0000000‭0 00000000 00000000 00000000 00000000 0010001|0 00000000 00000000 // b左移17位后的值(lb)
                                                    |<--这里后面的位补0

设:workerId  = 25,其(c)二进制左移12位计算过程如下:

             |<--这里开始左移12位    
‭00000000 0000|0000 00000000 00000000 00000000 00000000 00000000 00011001‬ // c = 25
00000000 00000000 00000000 00000000 00000000 00000001 1001|0000 00000000‬ // c左移12位后的值(lc)                                                                 
                                                          |<--这里后面的位补0

设:sequence = 0,其二进制如下:

00000000 00000000 00000000 00000000 00000000 00000000 0000‭0000 00000000‬ // sequence = 0

现在知道了每个部分左移后的值(la,lb,lc),代码可以简化成下面这样去理解:

return ((timestamp - 1288834974657) << 22) |
        (datacenterId << 17) |
        (workerId << 12) |
        sequence;
-----------------------------
           |
           |简化
          \|/
-----------------------------
return (la) |
        (lb) |
        (lc) |
        sequence;

上面的管道符号|在Java中也是一个位运算符。其含义是:
x的第n位和y的第n位 只要有一个是1,则结果的第n位也为1,否则为0,因此,我们对四个数的位或运算如下:

1  |                    41                        |  5  |   5  |     12      
    
   0|0001100 10100010 10111110 10001001 01011100 00|00000|0 0000|0000 00000000 //la
   0|000000‭0 00000000 00000000 00000000 00000000 00|10001|0 0000|0000 00000000 //lb
   0|0000000 00000000 00000000 00000000 00000000 00|00000|1 1001|0000 00000000 //lc
or 0|0000000 00000000 00000000 00000000 00000000 00|00000|0 0000|‭0000 00000000‬ //sequence
------------------------------------------------------------------------------------------
   0|0001100 10100010 10111110 10001001 01011100 00|10001|1 1001|‭0000 00000000‬ //结果:910499571847892992

结果计算过程:
1) 从至左列出1出现的下标(从0开始算):

0000  1   1   00  1   0  1  000  1   0  1  0  1  1  1  1  1  0 1   000 1 00 1  0 1  0   1  1  1  0000 1   000  1  1  1  00  1‭   0000 0000 0000
      59  58      55     53      49     47    45 44 43 42 41   39      35   32   30     28 27 26      21       17 16 15     12

2) 各个下标作为2的幂数来计算,并相加:

$ 2^{59}+2^{58}+2^{55}+2^{53}+2^{49}+2^{47}+2^{45}+2^{44}+2^{43}+
2^{42}+2^{41}+2^{39}+2^{35}+2^{32}+2^{30}+2^{28}+2^{27}+2^{26}+
2^{21}+2^{17}+2^{16}+2^{15}+2^{2} $
2^59}  : 576460752303423488
    2^58}  : 288230376151711744   
    2^55}  :  36028797018963968    
    2^53}  :   9007199254740992     
    2^49}  :    562949953421312      
    2^47}  :    140737488355328
    2^45}  :     35184372088832
    2^44}  :     17592186044416
    2^43}  :      8796093022208
    2^42}  :      4398046511104
    2^41}  :      2199023255552
    2^39}  :       549755813888
    2^35}  :        34359738368
    2^32}  :         4294967296
    2^30}  :         1073741824
    2^28}  :          268435456
    2^27}  :          134217728
    2^26}  :           67108864
    2^21}  :            2097152
    2^17}  :             131072
    2^16}  :              65536
    2^15}  :              32768
+   2^12}  :               4096
---------------------------------------- 
             910499571847892992

计算截图:
理解分布式id生成算法SnowFlake

跟测试程序打印出来的结果一样,手工验证完毕!

观察

1  |                    41                        |  5  |   5  |     12      
    
   0|0001100 10100010 10111110 10001001 01011100 00|     |      |              //la
   0|                                              |10001|      |              //lb
   0|                                              |     |1 1001|              //lc
or 0|                                              |     |      |‭0000 00000000‬ //sequence
------------------------------------------------------------------------------------------
   0|0001100 10100010 10111110 10001001 01011100 00|10001|1 1001|‭0000 00000000‬ //结果:910499571847892992

上面的64位我按1、41、5、5、12的位数截开了,方便观察。

  • 纵向观察发现:

    • 在41位那一段,除了la一行有值,其它行(lb、lc、sequence)都是0,(我爸其它)
    • 在左起第一个5位那一段,除了lb一行有值,其它行都是0
    • 在左起第二个5位那一段,除了lc一行有值,其它行都是0
    • 按照这规律,如果sequence是0以外的其它值,12位那段也会有值的,其它行都是0
  • 横向观察发现:

    • 在la行,由于左移了5+5+12位,5、5、12这三段都补0了,所以la行除了41那段外,其它肯定都是0
    • 同理,lb、lc、sequnece行也以此类推
    • 正因为左移的操作,使四个不同的值移到了SnowFlake理论上相应的位置,然后四行做位或运算(只要有1结果就是1),就把4段的二进制数合并成一个二进制数。

结论:
所以,在这段代码中

return ((timestamp - 1288834974657) << 22) |
        (datacenterId << 17) |
        (workerId << 12) |
        sequence;

左移运算是为了将数值移动到对应的段(41、5、5,12那段因为本来就在最右,因此不用左移)。

然后对每个左移后的值(la、lb、lc、sequence)做位或运算,是为了把各个短的数据合并起来,合并成一个二进制数。

最后转换成10进制,就是最终生成的id

扩展

在理解了这个算法之后,其实还有一些扩展的事情可以做:

  1. 根据自己业务修改每个位段存储的信息。算法是通用的,可以根据自己需求适当调整每段的大小以及存储的信息。
  2. 解密id,由于id的每段都保存了特定的信息,所以拿到一个id,应该可以尝试反推出原始的每个段的信息。反推出的信息可以帮助我们分析。比如作为订单,可以知道该订单的生成日期,负责处理的数据中心等等。

相关推荐