Serverless

Serverless trades server management for platform constraints—execution time limits, cold starts, stateless handlers, per-invocation billing, and vendor-specific integrations. It fits spiky workloads, event reactions, and small teams if observability and local dev ergonomics are solved.

On interviews: explain cold start mitigation, connection pooling limits, and step functions for workflows. Compare with containers on Kubernetes.

Common pitfalls: long-running tasks forced into functions; hidden costs from high-frequency invocations; debugging distributed traces across many small functions.

Checklist:

• Functions stay small and idempotent.
• Shared state externalized to managed stores.
• IaC defines triggers, IAM, and alarms.
• Load tests include cold-start behavior.