However server-side generation would mean that exchange rates are cached along with the page by Google and archive.org. An old version of the page would show old exchange rates.
I care a great deal about letting users consume the content even if the website goes down, either through archive.org, Google Cache, CloudFlare, Pocket or their own archives.
Particularly today when so many programmers accept re-implementation of systems primitives in higher level languages de jour, there exists a wry truth in the (admittedly somewhat condescending) adage:
Those who don't understand Unix are condemned to reinvent it, poorly. - Henry Spencer
This sounds like an old man with a hammer, reminiscing of times where every single problem was a nail. True, hammer works for the threaded wood screws somewhat, but the new engineers with their electric screwdrivers are more efficient and build sturdier things.
Even if you are hell-bent on doing stuff the Linux way why not propose a pair of Systemd service and timer? At least there will be logs. Cron is such an antiquated way of doing things.
We've found that NextJS's Incremental Static Regeneration[0] hits the sweet spot for when we have needed to do this. Specifying a regeneration time allows us to say 'this data can be at most X seconds old' but if thousands of users hit a URL at once only one request is made.
[0] https://nextjs.org/docs/basic-features/data-fetching/increme...