Nomad is massively underrated

I love Kubernetes—the ecosystem, the community, and what it has made possible for cloud-native infrastructure. I’ve helped clients adopt it through EKS and GKE.

While exploring how platforms like KodeKloud and KillerKoda provide ephemeral, terminal-based labs, I became interested in building something similar. Working on kompanion.sh led me to HashiCorp Nomad.

For this project, I needed to schedule a fairly predictable set of containers as jobs. Kubernetes was familiar, but running it myself meant taking on more configuration and maintenance than I wanted. Managed Kubernetes was difficult to justify for my budget. Nomad felt like a better fit for the scope of the problem. There are still operational responsibilities: the scheduler doesn’t make networking, isolation, upgrades, or cleanup disappear. Those belong in the decision too. What I appreciate about Nomad is that it made me reconsider a familiar default. Knowing Kubernetes well had made it my starting point. Building something of my own, with my own constraints and budget, gave me a reason to look further. I’m glad I did.