Rate limits protect your service from bursts, buggy clients and the occasional bad actor. The algorithm matters less than choosing the right key and returning a response clients can act on.
Pick the key first
- Per API key for authenticated traffic
- Per IP address only as a coarse outer layer
- Per route for expensive endpoints like search or exports
Token bucket in a few lines
limiter := rate.NewLimiter(rate.Limit(10), 20) // 10 per second, bursts of 20
func handler(w http.ResponseWriter, r *http.Request) {
if !limiter.Allow() {
w.Header().Set("Retry-After", "1")
http.Error(w, "too many requests", http.StatusTooManyRequests)
return
}
// ...
}
Always send 429 Too Many Requests with a Retry-After header. Well-behaved clients will back off on their own, and the rest will show up clearly in your metrics.

In a fleet of servers, a per-process limiter allows N times the intended rate. Share state, or divide the limit by the number of instances.
