How do serverless apps scale?

https://watch-aws-lambda-scale.com

I built this website to *show* how AWS Lambda scales in real-time.
Play with it and watch it scale in response to your requests. It's all live.

Watch AWS Lambda scale

Every gray bar is an AWS Lambda instance (kinda like a container).
Every orange block is an invocation: code runs to handle a request.

The Lambda app here just `sleep()` 0 to 1s. And it forwards events (start/end/etc.) via websocket to the React frontend (the visualization).

We can see that AWS Lambda boots new instances in real-time to handle more requests.
1 instance always handles maximum 1 request at a time, meaning more requests = more instances. KISS.
Costs: Lambda bills the orange part, not the gray parts (pay per use, per ms).
I expect <1$/mo to run this.
I set up alerts + scaling limits. If it's attacked continuously (unlikely, but I know some of you will ask), I can lower the limits to limit costs to $1/day.