Rate Limiter — Live

Overview

A distributed API rate limiter you can watch work. A Redis token-bucket limiter in NestJS, fronted by a live Next.js dashboard that visualises the bucket draining and requests flipping to HTTP 429 the instant traffic crosses the sustained rate.

How it works

Each request spends one token from a bucket that refills at a fixed rate — allowing a short burst while enforcing a steady long-run rate. The whole read-compute-write decision runs inside a single Redis Lua script, so concurrent requests across instances can't double-spend a token. Every decision is streamed to the browser over Server-Sent Events, which is what makes the bucket visible in real time.

// atomic in one round-trip — no lost updates under load
const [allowed, tokens] = await redis.tokenBucket(key, now, capacity, refill, 1);

Highlights

• Token-bucket algorithm with an atomic Redis Lua script (no double-spend under concurrency)
• Live SSE dashboard: token gauge, request stream, throughput chart, one-click traffic scenarios
• Standard X-RateLimit-* and Retry-After headers
• Proven three ways — unit (fake clock), integration (Testcontainers), and k6 load tests
• One docker compose up runs the whole stack; CI on GitHub Actions