DevOps is more Ops than Dev now

Tarasov Aleksandr
3 min readNov 1, 2019

Some weeks ago, I read the article “A developer goes to a DevOps conference” by Graham King and listened to the discussion about it in popular Russian language podcast Radio-T.

Photo by Piron Guillaume on Unsplash

“DevOps is Ops with new tools.”, Graham wrote.

This quote reflects very well my subjective perception of what the DevOps is gradually turning into.

I remember the time when I was working in the bank as a developer (in the division of innovations, to be more precise). There was traditional enterprise infrastructure, tools, and software delivery process. Still, we had carte blanche to change anything to grow up delivery speed dramatically. We took all edge technologies at the time like Spring Boot, Docker, and Ansible and built the brand new world to rule them all. It was a developer’s pipe dream. And now I understand that it was a scariest Ops nightmare.

We were young and brave and naive. Tools and approaches, in general, were easily understandable. Solutions were straightforward and dumb. We did not use Kubernetes, and we had no DevOps engineers (ok, we have no SRE as well). Docker, Ansible, some tricks for regenerating NGINX config — that was all. I believed that every engineer is happy to have freedom and responsibility (as Netflix culture said) to make his own decisions and be free to use the sharpest solution to solve the problem.

What do we see now? Kubernetes is the King. But how many developers try to deploy Kubernetes for production usage? Or even practice it on their local machine every day? Do they understand the difference between requests and limits and how they work under the hood to make the right choice for their app?

Besides Kubernetes, there are AWS or Google Cloud, Terraform, Helm, Prometheus, Fluentd, Jenkins, Spinnaker, GKE or something like Kuberspay and, in fact, tons of helpful and sometimes vital solutions.

As a Software Engineer, did you try to turn them on in a production environment and be successful?

The main problem at the moment that these solutions are complicated in setup, support, and usage. Moreover, they require specific knowledge and time to learn and practice to cook this soup right. You should touch them daily to stay on top and solve the engineering problems in the infrastructure field.

To be on edge, you should spend time on something that directly not help to implement business features or reduce tech debt. It something outside of development that ends up with a Docker container that will pass through a delivery pipeline. Ok, logs and monitoring could be a part of healthy developer concerns.

Who could spend much time to stay on top? System Administrators, DevOps Engineers, SRE Engineers, and YAML-Engineers who are typically in the operations field, not in development.

So, DevOps now is more Ops than Dev. Sad but true.

I believe that someday we see the day new Docker born.

--

--

Tarasov Aleksandr

Principal Platform Engineer @ Cenomi. All thoughts are my own.