What is serverless computing?

Serverless is an operating model more than a single product. Functions, managed queues, managed databases and API gateways can all be serverless. You do not choose or patch hosts, and capacity scales automatically with demand.

Characteristics

• Event-driven invocation from HTTP, queues, schedules, storage events or streams
• Automatic scaling, including scale to zero on some platforms
• Per-request or per-duration billing rather than paying for idle servers
• Stateless execution with external state in databases, caches or object storage
• Provider-managed runtime, patching and availability

export async function handler(event) {
  const body = JSON.parse(event.body);
  return {
    statusCode: 200,
    body: JSON.stringify({ orderId: body.orderId }),
  };
}

The trade-offs are execution time limits, memory and package size limits, cold starts, limited local disk, and concurrency quotas. Long-running batch jobs, custom kernels and latency-critical steady traffic often fit containers better.