It’s bewildering to me how Netlify, Vercel, Cloudflare and others have convinced the modern crop of developers this approach is fundamentally different than CDN caching and that “serverless” somehow doesn’t qualify as backend.
What Netlify (and Vercel) did is take all those disparate features across AWS that you use for static site hosting, and package them up with a nice UI and easy access.
What about end user experience?
It’s a CDN with a backend which is attractive. Effortless, infinitely scalable, dynamic web apps.
That’s awfully reminiscent of the MongoDB meme.
They're similarly pointy-clicky one-stop-shop for fairly complex external and internal networking, on high-tier plans, at least for Cloudflare.
I am very aware of the risks with things like this, but the appeal is clear.
It is a nice, managed solution for it where someone else takes care of it for you.
It doesn't need to be revolutionary to be worth paying for.
This will inevitably end with a whole lot of people not having a clue how to do what they think they know how to do when these companies go poof (or are swallowed by private equity).
Right now, the major players significantly abstract away from "how the web works" which means newer developers don't understand and are being led into a trap. Cute marketing terms like "serverless" confuse developers into thinking there is no server (or even worse, that there's such a thing as a "serverless developer").
I'm working on this problem right now: https://github.com/cheatcode/joystick. I know I sound like a curmudgeon when I say this stuff but it's a serious problem that will have dire consequences later (see: https://www.youtube.com/watch?v=ZSRHeXYDLko).
Well, that argument always goes. Linux users don't know how CPU scheduling works. CPU builders don't know how silicon etching works. The list is endless.
Web developers should build experiences with the best tools available. Apparently, these are the new preferred tools.
There is no trap.
Though in general our profession is built on and about building abstractions.
I don’t expect every 2010s web dev to know the intricacies of memory allocation, nor do I expect every 2020s web dev to know about nginx config or asset cache invalidation.
It’s handy sure, but required no. They’re all abstractions that you learn if and when necessary.
Hell, many don't know about tons of HTML features, many of which are literally decades old at this point. "Web" dev indeed.
Middle layers of software stacks are not implemented in a way that makes sense in isolation. There is always a reason why something is done a certain way and that reason usually comes from a lower layer.
For people who love to ask “why”, being forced to understand a single layer of a software stack is mental torture. Those who proclaim something is good because “you don’t have to think about x,y,z” are missing the point. Yes in the near term maybe, but long term its not the best solution.