Mastering System Resilience: Rate Limiting, Circuit Breaking, and Degradation

Ensuring high availability requires many measures, such as using Redis caching to reduce database load while handling cache penetration, avalanche, and breakdown issues. When facing sudden traffic spikes, three distributed architecture techniques—rate limiting, circuit breaking, and degradation—are essential.

Rate Limiting

Rate limiting is simple to understand: like the Forbidden City selling only 80,000 tickets per day, excess visitors are denied entry to avoid overcrowding and safety hazards. In software architecture, rate limiting allows only a portion of incoming traffic while rejecting the rest.

Common algorithms include counter method, sliding window, leaky bucket, and token bucket. The token bucket resembles the ticket analogy: tokens represent permission to access the service.

Rate limiting can also be applied per system or business flow; for example, when core system A is overloaded, calls from less critical system C can be limited while preserving access for important system B.

Circuit Breaking

In real life, a fuse acts as a circuit breaker, automatically disconnecting power during a short circuit to protect appliances. In distributed systems, a call chain A→B→C→D may suffer when downstream D reaches its capacity, causing slow responses and potential cascading failures.

Circuit breaking detects a surge of timeouts, indicating a failure, and stops further requests to the problematic service, breaking the chain. It can also monitor downstream health and gradually restore traffic once the service recovers.

Service Degradation

Degradation can be automated in code: when traffic surges, lower‑priority services are limited or disabled, preserving critical functionality. It can also be switched manually during peak events (e.g., Double 11 or 618) to disable non‑essential features, ensuring core services like shopping and payment remain operational.

In summary, rate limiting, circuit breaking, and degradation are key mechanisms to maintain system stability under excessive traffic.