限流，单机，分布式限流

矫情吗；* 2022-01-05 23:09 356阅读 0赞

总结：解决网站的高访问量，有三大利器，缓存，降级，限流，我这里讲解一下 常用的几种限流手段。

单机 ，有型号量，令牌通，漏桶，

分布式，可以考虑从网关，redis ngnix, 阿里 Sentinel 等手段解决 。话不多说代码如下：

1.单机：

1.1 型号量

// 线程池

ExecutorService exec = Executors.newCachedThreadPool();
    // 只能5个线程同时访问
    final Semaphore semp = new Semaphore(5);
    
    @Override
    public void limit() {
        
        // 模拟20个客户端访问
        for (int index = 0; index < 20; index++) {
        
            final int NO = index;
            Runnable run = new Runnable() {
        
                public void run() {
        
                    try {
        
                        // 获取许可 一直阻塞
                        semp.acquire();
                       // semp.tryAcquire(1, TimeUnit.MINUTES); //这个尝试 多长时间获取 到
                        //处理业务
                        log.info("Accessing: " + NO);
                        Thread.sleep((long) (Math.random() * 10000));
                        log.info("存在多少有效" + semp.availablePermits());
                    } catch (InterruptedException e) {
        
                        log.info("error:{}",e);
                    } finally {
        
                        // 访问完后，释放
                        semp.release();
                    }
                }
            };
            exec.execute(run);
        }
        // 退出线程池
        //exec.shutdown();
    }
    
    
    1.2 guava

ExecutorService exec = Executors.newCachedThreadPool();// 线程池
    final RateLimiter limit = RateLimiter.create(2);//1s只能有10个
    
    @Override
    public void limit() {
       
        // 模拟20个客户端访问
        for (int index = 0; index < 20; index++) {
       
            final int NO = index;
            Runnable run = new Runnable() {
       
                public void run() {
       
                    try {
       
                        // 获取许可
                        limit.acquire();
                        //处理业务
                        log.info("Accessing: " + NO);
                        Thread.sleep((long) (Math.random() * 10000));
                    } catch (InterruptedException e) {
       
                        log.info("error:{}", e);
                    }
                }
            };
            exec.submit(run);
        }
    }
    
    2.0 分布式版本

2.1  redis-spring-data

@Autowired
    private RedisTemplate redisTemplate;
    ExecutorService exec = Executors.newCachedThreadPool();// 线程池
    @Override
    public void limit() {
       
        //这里我用 包名+类名+方法
        String key = "com.example.demo.limit.redis.ResdisLimitServiceImpl.limit";
    
        RedisAtomicLong counter = new RedisAtomicLong(key, redisTemplate.getConnectionFactory());
        for (int i = 0; i < 100; i++) {
       
            final int no = i;
            exec.submit(new Runnable() {
       
                @Override
                public void run() {
       
                    if (counter.incrementAndGet() > 10) {
       
                        //拒绝请求
                        log.info("拒绝请求{}", no);
                    }
                    try {
       
                        //处理请求
                        log.info("处理请求{}", no);
                    } finally {
       
                        counter.decrementAndGet();
                    }
                    try {
       
                        Thread.sleep((long) (Math.random() * 10000));
                    } catch (InterruptedException e) {
       
    
                    }
                }
            });
        }
    }

2.2 redission 版本

@Autowired
    private RedissonClient redisson;
    
    ExecutorService exec = Executors.newCachedThreadPool();// 线程池
    
    @Override
    public void limit() {
       
        RSemaphore semaphore = redisson.getSemaphore("semaphore");
        semaphore.trySetPermits(10);
        log.info("有效：{}",semaphore.availablePermits());
        for (int i = 0; i <100 ; i++) {
       
            final int no = i;
            exec.submit(new Runnable() {
       
                @Override
                public void run() {
       
                    try {
       
                        if(semaphore.tryAcquire(1, TimeUnit.SECONDS)){
       
                            try {
       
                                log.info("处理业务{}",no);
                                Thread.sleep((long) (Math.random() * 10000));
                            }finally {
       
                                semaphore.release(1);
                            }
                        }else {
       
                            log.info("拒绝业务{}",no);
                        }
                    } catch (InterruptedException e) {
       
                        log.error("error:{}",e);
                    }
                }
            });
        }
    }

3.0 基于注解 redis开发 限流 使用方便 
    使用如下：
    
    @Limit(key = "test", period = 100, count = 10)

@PostMapping("dept/update")
    public ResponseResult updateDept(String name, Boolean leader, Integer id
                                    , Double age, BigDecimal price, ResponseResultCode em){
        
    
        return ResponseResultFactory.success("添加成功");
    }
    jmeter 测试如下 具体实现 参考我的代码库

转载于:https://www.cnblogs.com/lyc88/articles/11004120.html