System Design Interview
Edge rate limiting with token bucket and distributed counters.
Start with a single service applying token bucket logic.
What was missing: No separation of concerns, no redundancy, no async, no observability.
Why that's risky: Single point of failure, no protection from spikes, no monitoring.
What gets added: Nothing yet (MVP).
Trade-offs: Simple, but not production ready.
Introduce edge and API gateway for security and separation of concerns.
What was missing: No edge security, no API gateway, no separation of logic.
Why that's risky: Vulnerable to DDoS, no traffic shaping, logic not isolated.
What gets added: Edge, API gateway.
Trade-offs: Slightly more complex.
Add load balancer and multiple token bucket replicas for scale and redundancy.
What was missing: No redundancy, no horizontal scaling, single token bucket bottleneck.
Why that's risky: Single point of failure, cannot handle spikes.
What gets added: Load balancer, multiple token bucket replicas.
Trade-offs: More complex deployment.
Add distributed sync, async queue, and worker pool for scale and durability.
What was missing: No async sync, no worker pool, all updates synchronous.
Why that's risky: Spikes can overload token buckets, slow sync blocks requests.
What gets added: Async queue, worker pool, idempotency.
Trade-offs: More moving parts, eventual consistency for sync.
Add multi-region, monitoring/logging, and circuit breaker for resilience and observability.
What was missing: No multi-region, no monitoring, no circuit breaker, no observability.
Why that's risky: Failures can cascade, no alerting, no regional failover.
What gets added: Multi-region, monitoring/logging, circuit breaker.
Trade-offs: More moving parts, more operational complexity.
Add global quota sync, analytics DB, and feedback for advanced operations.
What was missing: No global quota sync, no analytics DB, no feedback loop.
Why that's risky: No global visibility, no analytics for tuning, no feedback for improvement.
What gets added: Global quota sync, analytics DB, feedback loop.
Trade-offs: More moving parts, more operational complexity.
Shard counters by user hash and use sliding windows with probabilistic sampling.
Use async queues to reconcile counter deltas and accept minor drift with periodic sync.
Edge placement blocks abusive traffic early and reduces downstream load.
The queue smooths counter updates and lets workers reconcile state without blocking requests.
You're in the middle of an interview session. Leaving now will end your current attempt.
Explore concept overviews, real-system examples, key tradeoffs, and interview talking points for each roadmap section.
You've conquered this phase. These are the skills you now own:
