Serverless Architecture
Run functions or containers on demand — no server management, auto-scale to zero, pay per invocation.
Interview tip Lead with a 30-second definition, then one real system example and name 2–3 designs where Serverless Architecture is non-negotiable.
① What it is (30 seconds)
Run functions or containers on demand — no server management, auto-scale to zero, pay per invocation.
② How it works in system design
API Gateway → Lambda/Cloud Functions. Cold start latency trade-off. State in external DB/S3. Good for bursty, event-triggered work. Limit execution time and memory.
Typical placement
Client→Edge / Gateway→Serverless→Services→Data stores
③ Concrete system design example
Scenario: Image thumbnail on upload: S3 event triggers Lambda, generates sizes, writes back to S3. Zero cost when no uploads; scales automatically on viral spike.
④ Important interview Q&A
| Question | Answer |
|---|---|
| Cold start problem? | Provisioned concurrency, smaller runtimes (Go/Rust), or warm pools for latency-sensitive paths. |
| When not serverless? | Long-running streams, WebSocket chat rooms, steady high QPS GPU inference. |
| Stateless rule? | All state in DynamoDB/S3; function instance is ephemeral. |
⑤ Seen in these system designs
- Instagram — thumbnail workers
- Notifications — email workers
- Google Photos Duplicate — batch jobs
In interviews, after explaining the concept, say: "This shows up directly in …" and link two designs.
⑥ Revision checklist
- Trigger source
- Cold start awareness
- State externalized
- Timeout limits
- Cost model