twitter snowflake: 零配置随机版

阅读更多

因为无法彻底解决系统时钟后调的问题, 而且嫌配置麻烦. 所以对snowflake做了一下修改:

1.消除了配置.

2.尽可能地消除时钟后调时的可用性问题.

3.生成对数据库索引友好的流水号.

 

具体实现:

/**
 * Snowflake java version with random string suffixed.
 * Translated  by [email protected] on 2017/3/22  from scala version at https://github.com/twitter/snowflake/releases/tag/snowflake-2010.
 */
class SnowflakeGenerator{
    private static final byte[] alphanumerics = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789".getBytes();

    /*2017-01-01 00:00:00*/
    private static final long twepoch = 1483200000000L;

    private static final long workerIdBits = 5L;
    /*31*/
    private static final long maxWorkerId = -1L ^ (-1L << workerIdBits);

    private static final long datacenterIdBits = 5L;
    /*31*/
    private static final long maxDatacenterId = -1L ^ (-1L << datacenterIdBits);

    private static final long sequenceBits = 12L;
    /*4095*/
    private static final long maxSequence = -1L ^ (-1L << sequenceBits);

    /*12*/
    private static final long workerIdShift = sequenceBits;
    /*17*/
    private static final long datacenterIdShift = sequenceBits + workerIdBits;
    /*22*/
    private static final long timestampLeftShift = sequenceBits + workerIdBits + datacenterIdBits;

    /*0-31*/
    private long workerId;
    /*0-31*/
    private long datacenterId;
    /*0-4095*/
    private long sequence = 0L;

    private long lastTimestamp = -1L;

    public SnowflakeGenerator() {
        this(randomLong(maxWorkerId), randomLong(maxDatacenterId));
    }

    public SnowflakeGenerator(long workerId, long datacenterId) {
        this(workerId, datacenterId, randomLong(maxSequence));
    }

    public SnowflakeGenerator(long workerId, long datacenterId, long sequence) {
        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));
        }
        this.workerId = workerId;
        this.datacenterId = datacenterId;
        this.sequence = sequence;
    }

    public String nextId() {
        return nextLongId() + createRandomAlphanumerics(7);
    }

    public synchronized long nextLongId() {
        long timestamp = System.currentTimeMillis();

        if (timestamp < lastTimestamp) {
            //allow clock moving backward since the result will be suffixed with random string
            //throw new IllegalStateException(String.format("Clock moved backwards.  Refusing to generate id for %d milliseconds", lastTimestamp - timestamp));
        }

        if (lastTimestamp == timestamp) {
            sequence = (sequence + 1) & maxSequence;
            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 = System.currentTimeMillis();
        while (timestamp <= lastTimestamp) {
            timestamp = System.currentTimeMillis();
        }
        return timestamp;
    }

    private static long randomLong(long maxValue) {
        return (long) (Math.random() * (maxValue + 1));
    }

    private String createRandomAlphanumerics(int length) {
        byte[] array = new byte[length];
        for (int i = 0; i < array.length; i++) {
            array[i] = (alphanumerics[(int) (Math.random() * alphanumerics.length)]);
        }
        return new String(array);
    }
}

 

解决的方法无外乎引入随机. 在同一秒的时间内, 出现碰撞的概率为150万万万亿分之一, 就是同一个人连续三次中大乐透一等奖的概率的30分之一. 如果公司有100万台服务器, 每个服务器上都运行着snowflake, 则在同一秒钟内出现碰撞的概率为1万万万亿分之一的数量级.

相对于官方snowflake算法而言, 引入随机无疑是一种倒退, 性能下降25%. 那也是没有办法的事情, 反正够用了.

 

相关阅读: 

twitter snowflake生成流水号的系统时间后调问题

twitter snowflake: kubernetes+docker+zookeeper实现workerId的自动分配