使用策略模式实现 Spring 分布式和单机限流

我们可以使用策略模式来统一单机限流和分布式限流的实现,提高代码的可扩展性和可维护性。

思路是定义一个 RateLimitStrategy 接口,并分别实现单机限流策略 LocalRateLimitStrategy 和分布式限流策略 DistributedRateLimitStrategy。在 AOP 切面中,根据配置决定使用哪种限流策略。

定义策略接口

public interface RateLimitStrategy {
    boolean tryAcquire(String key, double qps, long timeout, TimeUnit timeUnit);
}
1
2
3

实现单机限流策略

import com.google.common.util.concurrent.RateLimiter;

import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.TimeUnit;

public class LocalRateLimitStrategy implements RateLimitStrategy {

    private final Map<String, RateLimiter> rateLimiters = new ConcurrentHashMap<>();

    @Override
    public boolean tryAcquire(String key, double qps, long timeout, TimeUnit timeUnit) {
        RateLimiter limiter = rateLimiters.computeIfAbsent(key, k -> RateLimiter.create(qps));
        if (timeout > 0) {
            return limiter.tryAcquire(timeout, timeUnit);
        } else {
            return limiter.tryAcquire();
        }
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

实现分布式限流策略

import org.springframework.data.redis.core.RedisTemplate;
import org.springframework.data.redis.core.script.RedisScript;

import java.util.Collections;
import java.util.List;
import java.util.concurrent.TimeUnit;

public class DistributedRateLimitStrategy implements RateLimitStrategy {

    private final RedisTemplate<String, Object> redisTemplate;

    public DistributedRateLimitStrategy(RedisTemplate<String, Object> redisTemplate) {
        this.redisTemplate = redisTemplate;
    }

    @Override
    public boolean tryAcquire(String key, double qps, long timeout, TimeUnit timeUnit) {
        long window = timeUnit.toSeconds(timeout);
        List<String> keys = Collections.singletonList(key);

        String luaScript = buildLuaScript();
        RedisScript<Long> redisScript = new DefaultRedisScript<>(luaScript, Long.class);
        Long currentCount = redisTemplate.execute(redisScript, keys, Collections.singletonList(window), Collections.singletonList(qps));

        return currentCount <= qps;
    }

    private String buildLuaScript() {
       return "local key = KEYS[1]\n" +
                "local window = tonumber(ARGV[1])\n" +
                "local qps = tonumber(ARGV[2])\n" +
                "local current = redis.call('incrBy', key, 1)\n" +
                "if current == 1 then\n" +
                "    redis.call('expire', key, window)\n" +
                "end\n" +
                "if current > qps then\n" +
                "    return redis.call('decrBy', key, 1)\n" +
                "else\n" +
                "    return current\n" +
                "end";
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42

修改切面逻辑

import org.aspectj.lang.ProceedingJoinPoint;
import org.aspectj.lang.annotation.Around;
import org.aspectj.lang.annotation.Aspect;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.stereotype.Component;

import java.util.concurrent.TimeUnit;

@Aspect
@Component
public class RateLimitAspect {

    @Autowired
    private RateLimitStrategy rateLimitStrategy;

    @Around("@annotation(rateLimitAnnotation)")
    public Object around(ProceedingJoinPoint joinPoint, RateLimit rateLimitAnnotation) throws Throwable {
        String key = joinPoint.getSignature().toLongString();
        double qps = rateLimitAnnotation.qps();
        long timeout = rateLimitAnnotation.timeout();
        TimeUnit timeUnit = rateLimitAnnotation.timeUnit();

        boolean acquired = rateLimitStrategy.tryAcquire(key, qps, timeout, timeUnit);
        if (!acquired) {
            throw new RuntimeException("Rate limit exceeded");
        }

        return joinPoint.proceed();
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

在切面逻辑中,我们注入了 RateLimitStrategy 的实现类。根据配置决定使用单机限流还是分布式限流策略。

使用示例

@RestController
public class DemoController {

    @Autowired
    private RateLimitStrategy rateLimitStrategy;

    @GetMapping("/test")
    @ApiRateLimit(qps = 10, timeout = 60, timeUnit = TimeUnit.SECONDS)
    public String test() {
        return "hello world";
    }
}
1
2
3
4
5
6
7
8
9
10
11
12

在使用时,我们只需要在方法上标注 @RateLimit 注解即可,而不需要关心底层使用的是单机限流还是分布式限流。

配置限流策略

在 Spring 配置中,我们可以根据需求注入不同的 RateLimitStrategy 实现类:

// 单机限流配置
@Bean
public RateLimitStrategy localRateLimitStrategy() {
    return new LocalRateLimitStrategy();
}

// 分布式限流配置
@Bean
public RateLimitStrategy distributedRateLimitStrategy(RedisTemplate<String, Object> redisTemplate) {
    return new DistributedRateLimitStrategy(redisTemplate);
}
1
2
3
4
5
6
7
8
9
10
11

通过使用策略模式,我们将限流算法与具体的限流策略解耦,提高了代码的可扩展性和可维护性。未来如果需要新的限流策略,只需要实现 RateLimitStrategy 接口并配置即可,无需修改核心的限流逻辑。