Sure would have loved to see a breakdown of the breadth of functions and the nature of each function (did you simply just serve as a stripe proxy? did you squirrel away some data into a separate database? etc).

Hard to know just how much serverless really helped here as opposed to other solutions. Also, the cost of the serverless execution would be great to know (don't forget those API Gateway fees or other such "hidden" costs).

The tech empower benchmarks[0] w/ a single DB request and NodeJS show some pretty good results (10k-100k) and since the max was 2.5k. I think Node isn't a particularly performance conscious choice, though it does async well and that should be pretty common knowledge these days (that doing async well is a good idea for high load).

Also normally you'd use Cloudflare workers to serve some data very close to customers -- was that the use case here? or was it more just that the workers were a good fit for deployment ease? There's also the question of essentially subsidies -- I hear that cloudflare worker ingress (egress?) traffic is free right now?

[0]: https://www.techempower.com/benchmarks/#section=data-r19&hw=...