[xtracto]

> As developers who start writing big servers learn, scale matters. Crap you'd never think twice about when running a script on your workstation will bring your entire service down.

This reminded me of an issue we had at a previous startup that was growing really fast. There was a process that "created PDF invoices", which was coded by calling a (sync) API, which generated it on the fly.

The problem changed, once those PDF invoices became 100MB large, with hundreds of pages (required by business case). It's a completely different beast that the "MVP" developers did not thought about (as it is expected). Now you either code and maintain an async service which uploads to S3, along with the full lifecycle, or just buy a service to do it for you.

Scale definitely matters, and all systems change once you consider large scale data and workloads.

[dnautics]

wait did you use to work where I work? We fixed this, haha.

[xtracto]

I don't think so, but it is a recurring issue I've seen at several B2B startups. It makes sense if you think about it. Being B2B, customers' AP departments request detailed usage billing. Someone creates a simple endpoint to produce a PDF, and eventually PDFs get too large to handle synchronously.