What Is Rate Limiting?

Rate limiting restricts how many requests a client can make to an API within a given time window, protecting the service from overload or abuse.

Published November 17, 2026

Rate limiting caps how many requests a client can make in a time window, protecting your API from abuse and overload. Exceeding it returns HTTP 429 Too Many Requests.

The concept

Common algorithms: fixed window (100/min per key), sliding window and token bucket (tokens refill at a steady rate, allowing short bursts).

Limits are keyed by IP, API key or user. Responses should include headers like Retry-After and X-RateLimit-Remaining.

Token bucket

  1. Bucket holds up to N tokens, refilled steadily
  2. Each request consumes one token
  3. Empty bucket → reject with 429
  4. Tokens refill, requests resume

Practical examples

Laravel throttle

Route::middleware('throttle:60,1')->group(function () { /* 60 requests per minute */ });

Express

import rateLimit from 'express-rate-limit'
app.use(rateLimit({ windowMs: 60_000, max: 100 }))

429 response

HTTP/1.1 429 Too Many Requests
Retry-After: 30

Client back-off

if (res.status === 429) await sleep(Number(res.headers.get('Retry-After')) * 1000)

Common mistakes

  • Limiting only by IP behind shared NAT.
  • No Retry-After header.
  • Storing counters in process memory with multiple servers (use Redis).

Key takeaways

  • Cap requests per window.
  • 429 + Retry-After.
  • Share counters across servers.

FAQ

Rate limit vs throttle?

Often used interchangeably; throttling may slow requests instead of rejecting.

Where to enforce?

At the edge or gateway, plus sensitive endpoints in the app.

More HTTP articles