Hacker News Reader

Comment by runako | Hacker News Reader

No hate to any of the PaaSes out there, Vercel included. They truly serve a need.

That said, if you are an engineer planning on working in/around the field, I would strongly suggest developing some competence at basic Linux systems administration. (Also: learn SQL, even though it's out of fashion.)

Linux is probably the single technology where my knowledge has had the longest useful lifespan (SQL is probably second). There are Unix (System V) bits I learned decades ago that are still useful today, on Linux.

Then, you can use a PaaS if you want. But if it's not the right fit, you are in a position to do something else. You might find that designing your application with a modern compute stack (this is not a PaaS) gives you an unfair advantage.

View on HN

> learn SQL, even though it's out of fashion

In what world is SQL out of fashion??

There's now ~2 generations of professional engineers for whom SQL was rarely/never a thing to learn. Between the hard(er) split among front-end/back-end developers, ORMs improving, and the (flawed) idea that NoSQL would make SQL irrelevant, it has become somewhat of a niche skill.

Think about Firebase. One can be full stack on an app built on Firebase and be successful without ever touching SQL. Firebase is very popular, and has been for some time.

Source: I have worked with a set of otherwise solid engineering teams and can say that SQL familiarity has given me a leg up on very smart engineers who nonetheless do not do relational databases.

The only way you can get away with creating an application without touching sql is of you offload the logic to your backend language, and then I don't think you'd be efficient enough to scale.

Also can someone actually understand the logic of joins, indexes, pks, etc enough to create an efficientand scalable db, and not simply have learned sql by proximity?

> The only way you can get away with creating an application without touching sql

Please look at app platforms like Firebase[1]. There are absolutely complex Web applications running at scale that do not use SQL anywhere in the stack.

Aside from that, MongoDB and Redis are 17 years old; CouchDB is over 20. NoSQL is well-established at this point. All of the hyperscalers offer proprietary NoSQL databases, and have done so for years. A large number of developers uses those databases in production.

In our API-centric environment, there are a lot of apps that don't do much in the way of managing their own data directly at all, using mixtures APIs for auth and other key application functions.

> can someone actually understand the logic of joins, indexes, pks, etc enough to create an efficientand scalable db

If you are not using a relational database, these concepts do not necessarily apply.

1 - https://firebase.google.com/docs/firestore/query-data/querie...

I did some interviewing rounds with PaaS platforms for advisory roles. I loved Heroku and was ruined by Heroku so I thought maybe the industry has something new to offer.

The model is largely "built to be locked in" model. It is not something innovative. The issue is that to what scale of operation the platform considers there customer to be locked in where they can hand them a bill that compensates for the ease of entry model for all the hobby/free tier.

With Vercel I feel like these level is becoming lower and lower. You can within minutes launch a full startup with Vercel and AI assisted coding. And Vercel assumes that as long as you do not recieve any traffic that is good. The moment you recieve even a mild amount of traffic you are considered locked in.

To some people that is a fair trade because they have so little trust in their products in the first place investing hours instead minutes is a fair trade. If the traffic comes they are already in the green. If the traffic does not come any effort they have put in puts them in red. So, you put as little effort as possible to get thing out there.

I like Vercel. They have figured a monetization model for Slop SaaS. The other PaaS needs to catch up. In 2026, PaaS exists as a model to make revenue out of slop.

An easy way to host your own and fallback to a PaaS would eat the industry.

Reply on news.ycombinator.com