Skip to content
Devgo

Serverless

A cloud model where you run code without managing servers — the provider handles scaling, capacity, and maintenance. You pay for actual usage, not idle. Examples: AWS Lambda, Vercel, Cloudflare Workers. Great for sporadic traffic and event-driven apps. Not for everything.

When to use
  • Sporadic or variable traffic.
  • Event-driven workloads.
  • APIs and webhooks.
  • Teams without DevOps.
When not to
  • Constant high traffic (dedicated servers cheaper).
  • Long-running processes.
  • Stateful apps.
  • When you need full control.

Definition

Serverless means the cloud provider manages servers; you deploy functions or apps that scale to zero. Pay-per-use (invocations, duration, requests). No . Types: FaaS (Lambda, Workers) for functions, BaaS (Firestore, ) for managed backends, serverless containers (Fargate).

Why it exists

Managing servers (provisioning, scaling, patching) is overhead. Serverless lets developers focus on code while the provider handles infrastructure, scaling to zero when idle.

How it works

Deploy code ( or app). The provider runs it on demand, scales automatically, and charges per use. No idle costs (scales to zero). Triggers: , events, schedules. Examples: Lambda (functions), (full-stack), Workers (edge), Fargate (containers).

Scaling limits

Cold starts (latency on first call). Execution time limits. Vendor lock-in (each provider's runtime/API). Hard to debug. Not cost-effective for constant high traffic. State management between calls.

Cost shape

Pay per use (requests, duration). Free tiers. Cheap for sporadic traffic, expensive for constant load. No idle costs.

Alternatives

  • Containers (Docker/K8s) — More control, more management.
  • VPS (DigitalOcean) — Predictable cost, you manage scaling.

Failure modes

  • Cold starts.
  • Vendor lock-in.
  • Execution limits.
  • Hard to debug.
  • Unexpected costs at scale.

,