System Design Interview: Rate Limiter CheatSheet

This is a quick-reference companion for the full deep-dive post. Read the complete walkthrough here: System Design Interview: Full Guide.

Table of Contents

How to Use This CheatSheet in an Interview

• Start with requirements (2-3 minutes).
• Draw high-level architecture (5-7 minutes).
• Pick the hardest sub-problem and go deep.
• Close with scale, failures, and trade-offs.

One-Minute Requirements Checklist

• Clarify functional scope and APIs.
• Confirm latency, availability, and consistency targets.
• Ask for DAU/MAU, peak concurrency, and growth assumptions.
• Confirm security/compliance expectations.

Core Architecture Snapshot

flowchart TD
  A[Client] --> B[API Gateway]
  B --> C[Core Service]
  C --> D[Cache Layer]
  C --> E[Primary DB]
  C --> F[Queue/Stream]
  F --> G[Async Workers]

• Keep writes authoritative in the primary service/database.
• Use cache for read-heavy access paths.
• Push asynchronous work to queues/workers.

Critical Trade-Offs to Say Out Loud

• Latency vs consistency (eventual consistency is often acceptable for feeds/analytics).
• Simplicity vs flexibility (single store now, polyglot persistence later).
• Cost vs reliability (multi-region and replication increase spend).

Capacity Planning Quick Math

• Start from peak QPS, not average QPS.
• Compute write/read split and identify the dominant path.
• Add 2x headroom for spikes and failure events.
• Estimate storage for 1 day, 1 month, and 1 year.

Failure Cases to Mention

1. Region outage and failover strategy.
2. Queue backlog growth and retry/dead-letter handling.
3. Cache stampede and hot partition mitigation.
4. Partial writes and idempotency safeguards.

Rapid-Fire Follow-Up Answers

• How do you scale reads? Add cache + read replicas + CDN where applicable.
• How do you scale writes? Partition by a stable key and isolate hot shards.
• How do you keep data safe? Multi-AZ replication, backups, and restore drills.
• How do you handle abuse? AuthN/AuthZ, rate limiting, and anomaly detection.
• What would you improve next? Better observability and adaptive autoscaling.

Related Deep Dive

• Full post: System Design Interview: Full Guide