分布式场景全局唯一ID生成工具类(非雪花算法)

Posted 2022-11-22 伯约听风

这是一个分布式场景下全局唯一ID生成工具，类似于雪花算法(SnowFlake)。如果你需要生成订单号等类似前缀+yyMMddHHmmss+序列格式的全局唯一性序列可以使用。

结构

前缀yyMMddHHmmssworkerId序列ID

优点

• 与雪花算法一样，单机执行无需其他中间件做分布式协调
• 代码简单，可自行修改。雪花算法估计很多人还看不太明白吧

局限性

• 需要提前规划序号的最大值，最大值 > 预计单机一秒钟需要生成的序列最大值

最佳实践

通过ZK等维护WORKER_ID这样可以更好保证唯一性。

代码

public class BizSequenceGenarator 

	private static final String TIME_PATTERN = "yyMMddHHmmss";
	private static String[] paddingzeros = new String[]"","0","00","000","0000","00000","000000","0000000","00000000","000000000";
	private static final String WORKER_ID =  String.valueOf(RandomUtils.nextInt(100, 999));

	private final AtomicInteger NEXT_COUNTER;
	private String prefix;
	private int incrMin;
	private int incrMax;
	
	

	public BizSequenceGenarator(String prefix, int incrMax) 
		this.prefix = StringUtils.stripToNull(prefix);
		if(this.prefix != null)this.prefix = this.prefix.toUpperCase();
		this.incrMin = Integer.parseInt("1" + String.valueOf(incrMax).substring(1).replaceAll("[0-9]1", "0"));
		this.incrMax = incrMax;
		NEXT_COUNTER = new AtomicInteger(RandomUtils.nextInt(incrMin, incrMax/2));
	

	public String next() 
		StringBuilder sb = new StringBuilder();
		if(prefix != null)sb.append(prefix);
		sb.append(DateUtils.format(new Date(),TIME_PATTERN));
		sb.append(WORKER_ID);
		sb.append(buildIncrNumSequence());
		return sb.toString();
	

	private String buildIncrNumSequence() 
		int next = NEXT_COUNTER.incrementAndGet();
		if(incrMax - next < 5)
			next = NEXT_COUNTER.updateAndGet( (x) -> x >= incrMax ? RandomUtils.nextInt(incrMin, incrMax/2) : x + 1);
		
		String seq = String.valueOf(next);
		//补0
		int len = 5 - seq.length();
		if(len > 0)
			seq = paddingzeros[len] + seq;   
		
		return seq;
	
	
	public static void main(String[] args) 
		BizSequenceGenarator genarator = new BizSequenceGenarator("TEST", 9999);
		
		for (int i = 0; i < 10; i++) 
			System.out.println(genarator.next());