In the world of software, DevOps emerged as a necessary crutch; a way to bridge the gap between the code developers create, and the production deployment of that code. But here’s the reality: DevOps, as most of us know it, is temporary. It exists because the technology we rely on is incomplete, not because it’s the ideal solution. And NO, platform engineering is not the successor to DevOps, as much as marketing strategies would have you believe, they serve different purposes!
As the CEO of Portainer, I’ve seen firsthand how the complexity of modern application deployment has driven businesses to create entire teams dedicated to DevOps. Employing experts who specialize in containerizing applications; creating Dockerfiles, deployment manifests, Helm charts, and the deployment pipelines that get all of this running on Kubernetes. Why? Because other than a limited few cloud-hosted services like Vercel, there isn’t suitable technology that can handle these processes on its own. DevOps was born out of necessity, not out of technological readiness.
What DevOps Offers Today
Let’s break it down. Developers are forced to work closely with experts to navigate the intricacies of deploying containerized applications. As shown in the chart below, they rely on DevOps engineers to:
-
Create Dockerfiles and build images: Even the most seasoned developers often need assistance in creating these, ensuring compatibility and security. Even something as simple as the incorrect selection of a base image (full of CVEs and not maintained) can put your production application at risk!
-
Craft deployment manifests and Helm charts: Without precise specifications, deployment to Kubernetes is impossible, or at best unreliable. There is no escaping the specialist knowledge needed to lay out the application on the Kubernetes platform (load balancers, ingress controllers, secrets, configs, persistence, deployment modes, health checks, upgrade policies, etc).
- Support the deployment tooling: A container image and a deployment manifest does not equal a running application. You need to execute the deployment instructions against the environment, and that's where CD/GitOps tooling comes in. This is not simple and again requires specialist skills to set up and support.
-
Monitor and manage deployments on Kubernetes: Once deployed, applications require continuous oversight to ensure uptime, performance, and security across dynamic clusters. And yup, once again, specialist skill required... see a trend yet?
These are all complicated, specialized, time-intensive tasks; tasks that software should handle automatically.
Oh, and just looping back to my initial comment re Platform Engineering... well that covers the scope of the bottom right box. Now, I agree that a well-established Platform Engineering team likely wants to design and deploy an Internal Developer Portal, but for many, that is an aspiration vs a reality.
Flashback: What was the way
You dont need to go back in time very far to recall how applications used to be promoted into Production. Microsoft Visual Studio has a "compile and deploy" button, that handled the deployment to an IIS server. PHP developers used products like cPanel to FTP their works to a web/db server.. it was all very simple. But technology designed to improve the developer experience seems to have taken us backwards. The sheer amount of time developers spend NOT developing is insane. By some accounts, up to 30% of a developer's time (and more often that not, senior developers), is spent troubleshooting DevOps tasks. Imagine if that 30% could be reclaimed and put back into business value-adding tasks like feature development! This is what we need to get back to.. simplicity.
The Future: Moving Beyond DevOps
Imagine a world where a developer’s path to production is as as easy as it was? Writing code and pressing “deploy.” We’re working toward that, and platforms are slowly catching up. The technology isn’t there yet, but we want to deliver that. If software could automate these steps, enabling seamless code-to-production workflows without human intervention, would we still need DevOps?
Here’s where I think we’re headed: developers should be able to rely on intelligent systems that take care of the operational heavy lifting introduced by container based deployment, making the current need for DevOps expertise redundant. DevOps is a reaction to the technology gap, not a permanent fixture. As we evolve, we won’t need entire teams managing infrastructure intricacies; the platform will do it for us. Even better, the DevOps engineers we all have today can elevate themselves, focus on site reliability engineering, and work on increasing the SLA and performance of an application, not on "busy" tasks of operating a platform.
DevOps is a Stop-Gap
So, is DevOps the future? I believe it’s a stepping stone; one that will eventually fade as technology catches up. The end goal is simple: let developers focus on building great software and let intelligent platforms handle the complexities of deployment and management.
We should challenge ourselves to look beyond DevOps and invest in solutions that simplify the entire journey from code to production. That’s the future Portainer is working toward.
Neil
COMMENTS