Is Kubernetes a Full-Time Job?

William Konitzer
Mirantis
Published in
2 min readJan 24, 2022

Now, I prefer my language gender-neutral, but this quote from Kelsey Hightower, which I saw on a tweet from Container Registry, really struck home with me.

I remember the old days.

I remember when containers weren’t supposed to be a full-time job. Every operations person, and many developers, were supposed to know enough about running a Docker container to be able to use one, build one, and distribute them to others.

But it couldn’t last.

The reality was that knowledge of operating containers is specialized, and eventually most teams ended up with a ‘Docker Person’ who ended up helping everyone else to accomplish the trickier tasks.

At the same time that some jobs like database administrator were disappearing, merged with other responsibilities and often not handled by a full-time specialist; containers kept growing in importance and complexity.

With Kubernetes, it often feels like the work never stops. We may not be manually provisioning and restarting servers, but it feels like there’s always more to learn, more complexity to understand, and inevitably our ‘K8s specialist’ becomes a ‘K8s person’ who does nothing else.

Certainly Kubernetes is overkill for some jobs, but on the other hand managing Kubernetes and a few Helm charts takes a lot less work than 3–4 virtual machines, and at the same time is a lot more consistent/reproducible.

There’s one angle that I can’t help but mention: on a Reddit thread discussing this quote, a painfully honest commenter said:

In our business it’s actually a challenge to find one technology that will keep us busy for our entire careers. I hope Kubernetes is it. I hope I can retire on this and live out my career without worrying about being obsolete.

And really we can’t discount it: one of the reasons some of us as individual engineers are drawn to Kubernetes is because we know it’s too complex to just automate and fully forget about. Any organization that adopts a large distributed compute cluster is always going to have need of our services.

So what do you think, does Kubernetes mean Job Security? Will we all be doing kubectl in 10 years?

--

--