API Rate Limiting Cheat Sheet

Jump to a section:

• Gateway-level rate limiting
• Token bucket algorithm
• Leaky bucket algorithm
• Sliding window algorithm
• Distributed rate limiting
• User-based rate limiting
• API key rate limiting
• Custom rate limiting

Gateway-level rate limiting

• Gateway-level rate limiting is a popular approach to rate limiting that allows developers to set rate limits at the gateway level.
• Gateway-level rate limiting is typically implemented in API gateways such as Kong, Google's Apigee, or Amazon API Gateway.
• Gateway-level rate limiting can provide simple and effective rate limiting, but may not offer as much fine-grained control as other approaches.

Token bucket algorithm

image source

• The token bucket algorithm is a popular rate limiting algorithm that involves allocating tokens to API requests.
• The tokens are refilled at a set rate, and when an API request is made, it must consume a token.
• If there are no tokens available, the request is rejected.
• The token bucket algorithm is commonly used in many rate limiting libraries and tools, such as rate-limiter, redis-rate-limiter, and the Google Cloud Endpoints.

More: Token Bucket vs Bursty Rate Limiter by @animir

Leaky bucket algorithm

image source

• The leaky bucket algorithm is similar to the token bucket algorithm, but instead of allocating tokens, API requests are added to a "bucket" at a set rate.
• If the bucket overflows, the requests are rejected.
• The leaky bucket algorithm can be useful for smoothing out request bursts, and for ensuring that requests are processed at a consistent rate.

Sliding window algorithm

image source

• The sliding window algorithm is a rate limiting approach that involves tracking the number of requests made in a sliding window of time.
• If the number of requests exceeds a set limit, further requests are rejected.
• The sliding window algorithm is commonly used in many rate limiting libraries and tools, such as Django Ratelimit, Express Rate Limit, and the Kubernetes Rate Limiting.

More: Rate limiting using the Sliding Window algorithm by @satrobit

Distributed rate limiting

• For high-traffic APIs, it may be necessary to implement rate limiting across multiple servers.
• Distributed rate limiting algorithms such as Redis-based rate limiting or Consistent Hashing-based rate limiting can be used to implement rate limiting across multiple servers.
• Distributed rate limiting can help to ensure that rate limiting is consistent across multiple servers, and can help to reduce the impact of traffic spikes.

In this example, we'll create a simple Next.js application with a rate-limited API endpoint using Redis and Upstash. Upstash is a serverless Redis database provider that allows you to interact with Redis easily and cost-effectively.

First, let's create a new Next.js project:

npx create-next-app redis-rate-limit-examplecd redis-rate-limit-example

Install the required dependencies:

npm install [email protected] [email protected] [email protected]

Create a .env.local file in the project root to store your Upstash Redis credentials:

UPSTASH_REDIS_URL=your_upstash_redis_url_here

Replace your_upstash_redis_url_here with your actual Upstash Redis URL.

Create a new API route in pages/api/limited.js:

import { connectRedis } from '../../lib/redis';import rateLimit from 'express-rate-limit';import { createError } from 'micro';const redisClient = connectRedis();const rateLimiter = rateLimit({  store: new RedisStore({    client: redisClient,  }),  windowMs: 60 * 1000, // 1 minute  max: 5, // limit each IP to 5 requests per minute  handler: (req, res) => {    res.status(429).json({ message: 'Too many requests, please try again later.' });  },});export default async function handler(req, res) {  try {    await rateLimiter(req, res);  } catch (error) {    if (error instanceof createError.HttpError) {      return res.status(error.statusCode).json({ message: error.message });    }    res.status(500).json({ message: 'Internal server error' });  }  res.status(200).json({ message: 'Success! Your request was not rate-limited.' });}export const config = {  api: {    bodyParser: false,  },};

Create a lib/redis.js file to handle Redis connections:

import Redis from 'ioredis';let cachedRedis = null;export function connectRedis() {  if (cachedRedis) {    return cachedRedis;  }  const redis = new Redis(process.env.UPSTASH_REDIS_URL);  cachedRedis = redis;  return redis;}

Create a new RedisStore class in lib/redis-store.js:

import { connectRedis } from './redis';export class RedisStore {  constructor({ client } = {}) {    this.redis = client || connectRedis();  }  async get(key) {    const data = await this.redis.get(key);    return JSON.parse(data);  }  async set(key, value, ttl) {    await this.redis.set(key, JSON.stringify(value), 'EX', ttl);  }  async resetKey(key) {    await this.redis.del(key);  }}

Now you can test your rate-limited API endpoint by starting the development server:

npm run dev

Visit http://localhost:3000/api/limited in your browser or use a tool like Postman or curl to make requests. You should see the Success! Your request was not rate-limited. message. If you make more than 5 requests within a minute, you'll receive the rate limit message:

Too many requests, please try again later.

User-based rate limiting

• Some APIs may require rate limiting at the user level, rather than the IP address or client ID level.
• User-based rate limiting involves tracking the number of requests made by a particular user account, and limiting requests if the user exceeds a set limit.
• User-based rate limiting is commonly used in many API frameworks, such as Django Rest Framework, and can be implemented using session-based or token-based authentication.

API key rate limiting

• For APIs that require authentication with an API key, rate limiting can be implemented at the API key level.
• API key rate limiting involves tracking the number of requests made with a particular API key, and limiting requests if the key exceeds a set limit.
• API key rate limiting is commonly used in many API frameworks, such as Flask-Limiter, and can be implemented using API key-based authentication.

Custom rate limiting

• Finally, it's worth noting that there are many other rate limiting approaches that can be customized to suit the needs of a particular API.
• Some examples include adaptive rate limiting, which adjusts the rate limit based on the current traffic load, and request complexity-based rate limiting, which takes into account the complexity of individual requests when enforcing rate limits.
• Custom rate limiting approaches can be useful for optimizing the rate limiting strategy for a specific API use case.

For my latest project Pub Index API I am making use of an API gateway for rate-limiting.

More: RESTful API Design Cheatsheet