For more than a decade, the DevOps model has been a trending topic in the IT ecosystem. If you wanted to sound hip and modern, you talked about how DevOps would optimize the way software was delivered and managed, if only everyone would get on board with DevOps philosophies and tools. I’m writing this article to tell you that those days are over. However exactly you choose to define DevOps, the fact is that the DevOps model is no longer cutting-edge. In fact, I think DevOps has basically reached its end, and we’re transitioning into the post-DevOps age.
Here is why, and what that means for DevOps’s future.
What Does the End of DevOps Mean?
When I say that the DevOps model has reached its end or is over, I don’t mean that no one is practicing DevOps anymore. Obviously, that’s not the case. I mean instead that the changes that DevOps aimed to enact have been achieved, bringing DevOps as a movement to a conclusion.
In other words, DevOps has reached an equilibrium that has negated the significance and novelty of DevOps as a concept. You might call this an ending in the Fukuyaman sense.
Why DevOps Is Over
There are several reasons why DevOps is now over.
1. DevOps means everything and nothing.
The biggest reason why we’re at the end of DevOps is that DevOps has lost any sort of fixed meaning.
Today, DevOps has practically become synonymous with “modern IT.” Almost every software vendor under the sun markets its products or services as being made for DevOps in one way or another, despite the fact that many of these solutions existed before DevOps came along--and have not changed markedly since.
Just because your APM tool claims to provide “DevOps monitoring,” or your log aggregator promises “DevOps logging,” does not usually mean that the tools do anything in a DevOps-specific manner. They work the same way they did before DevOps became trendy; what has changed is the branding.
It wasn’t always this way. There was a time (it ended about five years ago) when the platforms that were marketed as DevOps actually catered specifically and directly to the sorts of things that you do as part of DevOps, like continuous integration or automated software deployments.
But, today, it feels like everything is DevOps. And when everything is DevOps, nothing is DevOps. That’s as good a reason as any to declare DevOps over, I think.
2. DevOps is pervasive.
DevOps has also reached its end in the sense that the concepts associated with DevOps--automation, continuous development and deployment, collaboration, and so on--are now widespread. In this sense, DevOps has little left to achieve.
Yes, there may be some opportunities to make DevOps practices a bit more “mature” in some organizations, as recent survey data suggests. There are also surely some organizations that are not yet “doing DevOps,” although the ambiguity of the meaning of DevOps makes it hard to enumerate its specific adoption rates. It’s not as if you either do DevOps or you don’t; most organizations have adopted it to one extent or another, even if few are doing DevOps in its entirety all the time.
This is to say that, by and large, most IT teams and developers have certainly heard of DevOps by now, and are embracing it in one sense or another. To help them get more of the value that DevOps is supposed to provide, the IT industry should move beyond the conversation about what DevOps is and why it matters. It should instead look for opportunities to achieve the next big innovation after DevOps.
3. DevOps has been subsumed by its offshoots.
DevOps has also been watered down almost beyond recognition by the fact that it has spawned so many other *Ops movements, such as QAOps and DevSecOps.
Most of these DevOps offshoots take DevOps philosophies and try to apply them to niches other than development or IT Ops. QAOps is about bringing DevOps ideas into Quality Assurance, for example, while DevSecOps tries to extend them to security.
That’s good and well, and there is value to pushing the boundaries of DevOps beyond software development and IT engineering. However, I think we’ve reached the point where DevOps has generated so many secondary concepts that they call into question the point of DevOps itself. We’ve created a world where DevOps-proper seems outdated and boring.
That’s another reason why I think DevOps is at its end, or that we’ve at least reached what you might call a post-DevOps age. DevOps has become more of an historical reference point for IT modernization than it is modernization itself.
4. DevOps never became a discipline.
For a time, it seemed like DevOps might turn into a discipline of its own. Instead of being trained as programmers or IT engineers, everyone would learn the ways of DevOps, and practice those first and foremost.
But that hasn’t really happened. There are certainly many DevOps job titles out there, but many of the positions look merely like traditional developer or IT engineer roles with the DevOps label ironed on. More to the point, DevOps hasn’t become a real specialization. There are a handful of DevOps certifications out there, and some DevOps training programs, but you’ll be hard-pressed to find a university that offers a degree in DevOps, or a large community of practitioners who think of their professional identities as being DevOps and only DevOps.
Compare DevOps in this sense to actual disciplines, like programming, IT security or network engineering, and DevOps’s lack of enduring substance becomes clear. These disciplines will be around for decades, and are bricked into the way that the IT industry trains and structures work. I wouldn’t bet on the same being true of DevOps.
What the Post-DevOps Age Looks Like
Again, the fact that DevOps is over doesn’t mean that DevOps no longer matters. On the contrary: We’ve now entered a post-DevOps age in which DevOps will remain the basic frame of reference for software delivery and management, even if it ceases to be the be-all, end-all of the IT industry that it has been for the past decade.
You could compare DevOps in this respect to virtualization. Twenty years ago, virtualization à la VMware was a huge deal. Then, during the past decade, virtual machines started to look dated--but they did not disappear. Instead, they became the reference point against which newer innovations (like containers) are measured. You might say we live in a post-virtualization age--not in the sense that virtual machines have disappeared (they certainly haven’t), but because we have evolved past the time when virtualization was a trailblazing way to build your infrastructure.
I suspect, too, that the end of DevOps as an innovative force will mean the end of the weight carried by associated concepts, such as cloud-native computing, and the vogue surrounding microservices, containers, Kubernetes and the like. These things, too, won’t go away, but they will lose their steam and recede conceptually into the background as newer innovations take hold.
DevOps hasn’t gone away, exactly, but it has ceased to be the innovative, disruptive force it once was. I’m not sure where the next big thing in IT will emerge (though I have my guesses), but I am confident it won’t be within the niche surrounding DevOps.