What is serverless computing?
Asked 11d agoby IT-QA1 answer
glossaryserverlessclouddevops
What does 'serverless' actually mean — there's obviously still a server somewhere?
1 Answer
- AIIT-QA Assistant11d ago*AI-drafted answer — reviewed for correctness. Please verify against your own environment and versions; corrections and better answers are welcome below.* Right — 'serverless' doesn't mean no servers; it means *you* don't provision or manage them. You deploy code (functions) or use managed services, and the cloud provider runs, scales, and bills them on demand. The server management is abstracted away. The defining traits: 1. **Scale to zero**: when there's no traffic, nothing runs and you pay nothing. It scales up automatically with load. 2. **Pay per use**: billed for actual execution time/requests, not for idle capacity. 3. **No infrastructure management**: no patching, capacity planning, or load balancers to run. Canonical examples: AWS Lambda / Cloud Functions (Functions-as-a-Service), plus managed data services (DynamoDB, Firestore) and platforms like Cloud Run. Trade-offs to know: - **Cold starts**: an idle function must spin up on the first request, adding latency. - **Statelessness**: functions are ephemeral, so state must live in external stores. - **Execution limits**: time and memory caps make it a poor fit for long-running jobs. - **Cost at high, steady scale**: per-request pricing can exceed a plain always-on server once traffic is large and constant. It shines for spiky, event-driven, or low/variable-traffic workloads where paying nothing at idle is the win.