There was a time when backend engineers were the backbone of software development. They built APIs, managed databases, optimized server performance...
For further actions, you may consider blocking this person and/or reporting abuse
Software development is always changing.
But who says backend developers can't change? And haven't changed already.
I have seen websites change from static to dynamic. With dynamic meaning adding a database to the server the webserver runs on. Now with cloud platforms servers are running in network clusters.
If you think serverless is the thing that is the silver bullet, you buy into the hype. And that is never a good thing.
I see serverless as an extra tool that has its place in the development environment.
It seems that you don't understand all the decisions that go into backend development. And you start from a limited tool to pile on more work to positions that have their own speciality.
Devops in my opinion is an operations job, instead of configuring servers manually now it can be done with code. Devops don't touch application code, if they don't need to.
It is not that when someone specialises in one aspect of software development, that they forget all the other parts. For a backend developers that is impossible because that position is in the middle of frontend and operations.
I'm more worried for you, than for backend developers.
You already called yourself out. Dynamic sites have been around since the 90's and static sites have been they hype. Serverless has been around over a decade strong. The article is straight facts. Backend engineering is being replaced by AI first and right now. The real problem is terrible front-end that skirts backend dedication and foresight.
With static sites I'm not referring to the jamstack version, if that is the hype you mention.
Things like geocities where dynamic, otherwise there was no way people could add their own content. I knew there where java applets that brought dynamic parts. I knew asp, without the .net, but it was expensive to host a server. And then came php, and things got cheap enough to build your own dynamic website experiments.
I'm not saying it is a bad thing. It isn't the only way forward.
The post does not mention AI as the cause of the diminishing backend role.
AI at this time can't do reasoning. You can't feed it an idea and expect it to come up with a codebase on its own.
You have to spell it out with all the context you can provide, before it comes close. And then you have to verify if that is what you wanted.
Even in an AI world developers matter.
I'm not sure if you are ranting about frontend development or AI?
And it also feels like you are contradicting yourself. Because first you agree with the article, but here you agree with the sentiment of my comment.
You tried to touch many points about serverless architecture but you forgot very critical disadvantage of serverless which is cost. Do you know how costly it becomes when your serverless usage increases? I don't think you have full picture of what you are talking about.
My company opted out of serverless architectures because of its increasing costs
Good decision by your company. The traditional backend is so cheap and reliable as well.
It's great point. If you have more users concurrently requesting, serverless cost is going to be very high!!!!
Who has build those 'serverless' services?
Guess??
Backend engineers
I'd say there's definitely a kind of a shift (thanks to hyperscalers). I usually describe it as libraries are evolving into online services.
Say, if years ago you'd look at libs as pieces of functionality that you can combine - now you have more and more ways of combining entire services instead. Anyway, it's a long story, won't go deeper in this comment.
Specifically about cloud functions though - yes, they are pretty good especially when you are starting something new - very quick to develop and deploy. But I usually recommend watching them a bit more closely - one day you will grow to a level of demand (requests) that's worth handling in a more 'tightly packed' way. That's when you might prefer building your own backends and spin them in containers or something...
A bit like - the moment you get 24/7 demand for a CPU core - get a CPU core and fill it with work...
Good writeup, but I think the emphasis on serverless might be slightly overrated here. I think it is worth considering the broader shift in what it means to be a software engineer today. The industry isn’t just changing tools; it is redefining the role of engineers themselves.
Modern engineers are increasingly valued as problem solvers first and coders second. The explosion of abstractions (like serverless, low-code platforms, and managed services) means engineers no longer need to obsess over infrastructure minutiae or syntax quirks. Instead, they’re tasked with asking: How do I solve this business problem most effectively?
This requires a mindset shift:
Serverless is a tool, not a philosophy. The true evolution is engineers becoming strategic decision-makers who optimize for value, not just lines of code.
One problem with your argument, isn't all these serverless platforms essential back ends? Which means a backed developer created your "backend". The problem isn't backend jobs/code is disappearing it's the jobs and code are shifting, so you don't see them in your world but the world is a bigger place than you think.
Interesting. I find the reverse. Front end jobs are being demoted in salary by nearly half and competition for the role is steeper due to everyone and their dog learning basic React. Never mind that they lack html semantics, accessibility, mobile responsive structure, patient dogwork in performance optimising, smooth animation, transitions, good ui experience.
Yes the environment has changed but I think its a combination of austerity and the advent of ai to answer questions and fill in the knowledge gaps.
Regarding companies just using lambda and heaven forbid dynamodb in a small scale, have you seen the cost for running that and the benchmarks? Having spent the last 4 months playing with AWS exlusively I can confidently say its just a different skill, much like azure and gcp.
The benchmarks I've read online and cumulative cost under loas all steer more positively towards ec2 and ecs over complete lambda serverless architecture. Also it compartmentalises the dependencies for any migration away from depending on AWS, should their pricing model or company fail to be fair.
I love fiddling with servers 😊 but I also use serverless tech wherever it makes sense to do so.
The tech world is big enough to specialize in whatever you want to, so long as there is a good enough demand for that skill. With the amount of Linux servers out there in the wild, there are plenty of jobs. If you work hard enough and keep improving your skill set, there will always be opportunities.
There is no "my way is the right way". This is the problem with the internet in recent years, the fanboyism.
I do agree with "Adapt or Get Left Behind"; from day one as a software engineer you have to understand that tech changes often, the underlying engineering not so much but the APIs, tooling, and frameworks do. You must keep up, you don't have to know everything, pick a specialization and just aim for mastery.
If the tech you picked dies off, you know how to learn, so just learn the next thing. Not that hard.
You missed comparing cost factor at scale. Also let's say if you have to work at these tech giants,underlying infra would still be kubernetes.
Serverless or managed services = you are paying for the convenience.
It’s just a tool. Not your silver bullet.
Do you think serverless is the silver bullet. As number of users grow sleeveless cost shoot up as well, as opposed traditional system. Heck even most of companies won't even migrated fully to cloud.
But what about WebSockets?
There are some tools that tries got the best of both worlds (classic vs cloud), such as knative for build "lambda" functions whiteout too much costs.
serverless basically adhere to this motto of ship fast break things, while traditional more like cheaper but complicated, but i do love the traditional one
Happy to see more serverless geeks!
No mention of a clear path or suggestions of how to learn these skills in a way that is relevant to companies hiring for them in an attempt to migrate to cloud soa.
With serveless you must wake up at midnight to check your email, just to confirm you dont have a bill that looks like you were training a model.