Elevate your skills with unrestricted access to system design courses—your path to interview success and advanced engineering knowledge.

Everything in Free, plus

Pro Yearly

When Redis is unavailable, allow all requests through without rate limiting

Fail-Open (Allow All)

When Redis is unavailable, deny all requests as if rate limited

Fail-Closed (Deny All)

When Redis is unavailable, fall back to local in-memory rate limiting per gateway

Local Fallback (Coarse Limiting)

Redis Failure Handling Strategy

Enforce limits inline at the API gateway before requests reach backend services

Gateway/Edge (Recommended)

Gateway calls a separate rate limiter service to get allow/deny decisions

Dedicated Service

Each service (or sidecar) enforces limits locally using in-process state

In-Process / Sidecar

Rate Limiter Placement Options

Design a distributed rate limiter to protect API availability and prevent abuse

Gateway Placement with Inline Filter

Gateway-Limiter Interface

Identify the Requestor

Burst-Tolerant Algorithm

Consistent Enforcement Across Gateways

Policy/Config Management

This table summarizes what interviewers typically expect at each seniority level for rate limiter design.


Level Expectations

Design Rate Limiter | System Design Interview Question

Your API runs in 5 AWS regions. A tenant with a global limit of 10,000 requests per hour sends traffic to all regions. How do you prevent them from exceeding their quota by 5× (2,000 requests per region)?

Achieving Global Fairness Across Regions

Different API endpoints have vastly different backend costs. A search query costs 100ms of CPU and hits 3 databases. A user profile lookup costs 5ms and hits cache. Should they count equally toward rate limits?

Cost-Based vs Count-Based Rate Limiting

You're building a batch processing API where users submit jobs that consume quota over time (e.g., 'process 1M records' consumes 1M from daily quota). How do you prevent double-spending when requests are distributed across gateways?

Distributed Quota Reservation System

The requirements state 100% accuracy at 1M TPS. But strict atomicity limits Redis throughput. What if we relaxed accuracy to 99%? How would that change the architecture?

Challenging Constraints: Accuracy vs Throughput

Your API has 80% unused capacity at night but hits limits during business hours. How would you design a system that encourages more usage during off-peak while protecting peak capacity?

Soft Limits and Dynamic Quotas

Design Rate Limiter

Non-Functional Requirements

API (optional)

The rate limiter must run inline at the API gateway to block excess traffic before it reaches backend services.


High Level Design

The gateway must call a limiter module and react correctly on allow vs deny, including returning proper HTTP responses to the client.


The system must determine a stable identity per request (user, API key, tenant, IP) to track usage fairly and prevent abuse.


The limiter must allow short bursts of requests while enforcing a long-term average rate per identity.


The limiter must enforce limits correctly across multiple API gateway instances, so traffic cannot bypass limits by hitting different instances.


Operators must be able to update rate limit policies safely without redeploying gateways.


Design Rate Limiter

Practice Mode Coming Soon