通常为了防止突然过多请求或系统处理能力有限时，为了保护下游接口，通常会对下游接口限流，下面说一下使用redis实现简单限流。
可以使用zset对象实现：
- key：为对象的行为，即限流器的名字标识
- score：用时间戳来实现计时，当前时间-窗口时长 即为当前时间段开始时间，结束时间就为当前时间
- value：能体现一次唯一请求即可，但是要注意节省内存
下面看下Redisson的Redisson的RedissonRateLimiter基于rediss的zset是如何限流的：
1. 创建RedissonRateLimiter
使用RedissonClient客户端创建限流器:RRateLimiter rateLimiter = this.redissonClient.getRateLimiter(rateLimiterName);
初始化限流器的同时就给限流器起了名字，及对象的行为，如myRateLimit，下面用到限流器名的地名军用myRateLimit代替
# org.redisson.Redisson
//name:限流器名
@Override
public RRateLimiter getRateLimiter(String name) {
return new RedissonRateLimiter(commandExecutor, name);
}
2. 配置限流器
创建限流器后要对其进行配置，并将配置存到redis总，主要配置包括
- RateType：限流器类型：
- OVERALL:分布式限流;
- PER_CLIENT:单例限流
- rate：速率
- rateInterval： 窗口时长
- RateIntervalUnit ：时间单位，如毫秒
# org.redisson.RedissonRateLimiter
boolean trySetRate(RateType mode, long rate, long rateInterval, RateIntervalUnit rateIntervalUnit);
//将限流器配置设置到redis的hash中
@Override
public RFuture<Void> setRateAsync(RateType type, long rate, long rateInterval, RateIntervalUnit unit) {
return commandExecutor.evalWriteAsync(getRawName(), LongCodec.INSTANCE, RedisCommands.EVAL_BOOLEAN,
// myRateLimt->rate:rateInterval内最多多少permits
"redis.call('hset', KEYS[1], 'rate', ARGV[1]);"
//myRateLimt->interval：毫秒数
+ "redis.call('hset', KEYS[1], 'interval', ARGV[2]);"
//myRateLimt->0/1
+ "redis.call('hset', KEYS[1], 'type', ARGV[3]);"
//删除{myRateLimt}:value->{myRateLimt}:permits
+ "redis.call('del', KEYS[2], KEYS[3]);",
//(限流器名：myRateLimit ，{myRateLimt}:value ,{myRateLimt}:permits),rateInterval内最多多少permits,毫秒数,RateType在Enum位置
Arrays.asList(getRawName(), getValueName(), getPermitsName()), rate, unit.toMilis(rateInterval), type.ordinal())
3. 申请令牌
# org.redisson.RedissonRateLimiter
// permitsName:zset，key={myRateLimit}:permits,score为存放时的当前时间戳，value为(随机数+请求量)字符串
// valueName:string,key={myRateLimit}:value,当前窗口内期间剩余可请求量
private <T> RFuture<T> tryAcquireAsync(RedisCommand<T> command, Long value) {
return commandExecutor.evalWriteAsync(getRawName(), LongCodec.INSTANCE, command,
//1.从redis中获取限流器配置：rate,interval,type，获取不到报错
"local rate = redis.call('hget', KEYS[1], 'rate');"
+ "local interval = redis.call('hget', KEYS[1], 'interval');"
+ "local type = redis.call('hget', KEYS[1], 'type');"
+ "assert(rate ~= false and interval ~= false and type ~= false, 'RateLimiter is not initialized')"
//2.若type为PER_CLIENT类型，使用基于client的，key带有uuid
//1为单例，即不同实例的相同限流器名使用各实例的限流器;对应于服务相当于同一实例使用同一限流器，
//0为全局分布式，相同的限流器名在使用同一个限流器,对应于服务相当于同一服务使用同一限流器
//目前业务大多使用的是0
+ "local valueName = KEYS[2];"
+ "local permitsName = KEYS[4];"
+ "if type == '1' then "
+ "valueName = KEYS[3];"
+ "permitsName = KEYS[5];"
+ "end;"
//3.rate < 此次请求的permits，报错
+ "assert(tonumber(rate) >= tonumber(ARGV[1]), 'Requested permits amount could not exceed defined rate'); "
//4.获取{myRateLimit}:value值
+ "local currentValue = redis.call('get', valueName); "
+ "if currentValue ~= false then "
//5.1 不是第一次请求
//5.1.1 获取已过期的请求数{myRateLimit}:permits，过期定义为：当前时间-interval
+ "local expiredValues = redis.call('zrangebyscore', permitsName, 0, tonumber(ARGV[2]) - interval); "
+ "local released = 0; "
+ "for i, v in ipairs(expiredValues) do "
+ "local random, permits = struct.unpack('fI', v);"
+ "released = released + permits;"
+ "end; "
//5.1.2 过期请求数 >0,说明之前有请求量，删除过期的请求，即为节省内存，只保留窗口内的记录，重设{myRateLimit}:value。。。
+ "if released > 0 then "
+ "redis.call('zremrangebyscore', permitsName, 0, tonumber(ARGV[2]) - interval); "
+ "if tonumber(currentValue) + released > tonumber(rate) then "
+ "currentValue = tonumber(rate) - redis.call('zcard', permitsName); "
+ "else "
+ "currentValue = tonumber(currentValue) + released; "
+ "end; "
+ "redis.call('set', valueName, currentValue);"
+ "end;"
+ "if tonumber(currentValue) < tonumber(ARGV[1]) then "
//5.1.3 本次被限流，返回剩余过期时间delay，即如果时block模式，delay后重试
+ "local firstValue = redis.call('zrange', permitsName, 0, 0, 'withscores'); "
+ "return 3 + interval - (tonumber(ARGV[2]) - tonumber(firstValue[2]));"
+ "else "
//5.1.4 本次可以请求，本次请求添加到{myRateLimit}:permits，剩余可请求量{myRateLimit}:value -1,返回null
+ "redis.call('zadd', permitsName, ARGV[2], struct.pack('fI', ARGV[3], ARGV[1])); "
+ "redis.call('decrby', valueName, ARGV[1]); "
+ "return nil; "
+ "end; "
+ "else "
//5.2 即第一次请求，返回null
//设置String类型key{myRateLimit}:value为rate，
//往zset中添加key为{myRateLimit}:permits，score为当前时间戳,value为(随机数转成float,ratez转成Integer之后的字符数)
// 将{myRateLimit}:value 将去此次请求量permits
+ "redis.call('set', valueName, rate); "
+ "redis.call('zadd', permitsName, ARGV[2], struct.pack('fI', ARGV[3], ARGV[1])); "
+ "redis.call('decrby', valueName, ARGV[1]); "
+ "return nil; "
+ "end;",
//(myRateLimit,{myRateLimit}:value,{myRateLimit}:uuid,{myRateLimit}:permits,{{myRateLimit}:permits}:uuid
Arrays.asList(getRawName(), getValueName(), getClientValueName(), getPermitsName(), getClientPermitsName()),
//
value, System.currentTimeMillis(), ThreadLocalRandom.current().nextLong());
}
- 限流器配置存到hash结构的redis中，key为限流器名myRateLimt
- 时间段即窗口利用zset实现，score为请求时时间戳，这样就可以实现时间计算了，value为单次请求的令牌数
- 为了节约内存，将窗口外的数据删除，即当前时间戳-窗口时间段interval
4. 限流后处理
# org.redisson.RedissonRateLimiter
private void tryAcquireAsync(long permits, RPromise<Boolean> promise, long timeoutInMillis) {
long s = System.currentTimeMillis();
//1.请求令牌
RFuture<Long> future = tryAcquireAsync(RedisCommands.EVAL_LONG, permits);
future.onComplete((delay, e) -> {
if (e != null) {
//2.请求异常直接返回失败
promise.tryFailure(e);
return;
}
if (delay == null) {
//3.令牌获取成功返回
promise.trySuccess(true);
return;
}
if (timeoutInMillis == -1) {
//4.令牌获取时超过限制，且RedissonRateLimiter.Type为BLOCK，delay秒后重试
commandExecutor.getConnectionManager().getGroup().schedule(() -> {
tryAcquireAsync(permits, promise, timeoutInMillis);
}, delay, TimeUnit.MILLISECONDS);
return;
}
//5.timeout=执行开始到返回的时间
//令牌获取时超过限制，且RedissonRateLimiter.Type为TRY,在未timeout之前且在窗口内重试，
//否则返回失败
long el = System.currentTimeMillis() - s;
long remains = timeoutInMillis - el;
if (remains <= 0) {
promise.trySuccess(false);
return;
}
if (remains < delay) {
commandExecutor.getConnectionManager().getGroup().schedule(() -> {
promise.trySuccess(false);
}, remains, TimeUnit.MILLISECONDS);
} else {
long start = System.currentTimeMillis();
commandExecutor.getConnectionManager().getGroup().schedule(() -> {
long elapsed = System.currentTimeMillis() - start;
if (remains <= elapsed) {
promise.trySuccess(false);
return;
}
tryAcquireAsync(permits, promise, remains - elapsed);
}, delay, TimeUnit.MILLISECONDS);
}
});
}
3386
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
