Cold starts

The first request to a new function or container instance pays for loading the runtime and code, from about 100 ms in Go to around a second in Java. Provisioned capacity keeps instances warm.

Where it shows up in real systems

  • Serverless function

    There is no server to size: capacity is the concurrency limit. When traffic rises faster than warm instances exist, requests pay a cold start, which is much longer for Java than for Go. Very cheap at low volume, expensive at a steady high volume.

    In production: AWS Lambda, Azure Functions, Google Cloud Functions, Cloudflare Workers

  • Edge function

    Ideal for redirects, A/B tests, authentication checks and personalisation without a round trip to your region. Execution time and memory are tightly limited, and anything that needs your database still pays the trip back to it.

    In production: Cloudflare Workers, AWS Lambda@Edge, CloudFront Functions, Vercel Edge Functions, Fastly Compute

See it break, then fix it

Build an architecture, send it a day of traffic and outages, and learn why it holds or fails. Free to start.

Open the learning path Open the sandbox