Hey everyone! My company is currently using Red Hat OpenShift, but we're finding the licensing costs to be quite high, especially as we scale our VM and BareMetal usage. We're looking into alternatives and planning a proof of concept (PoC) to find a more stable, cost-effective, and simpler Kubernetes solution. One of our goals is to use this PoC as leverage in our upcoming contract renewal discussions with Red Hat.
Right now, I'm considering two main scenarios:
1. **OKD (Community OpenShift):** This is the technically easiest option and would require minimal disruption for our teams. However, I'm concerned about the project's true independence and the potential lingering reliance on the Red Hat ecosystem.
2. **Talos Linux + Omni (or not):** This approach appeals to me for its highly secure, pure Kubernetes style. I like the idea of an immutable operating system focused on APIs and no SSH, freeing our teams from traditional OS management burdens.
I'd love to connect with anyone who has migrated from OpenShift/OKD to a pure Kubernetes setup, especially if you've used Talos.
I'd particularly like to hear about:
- Migration challenges: How tough was it to convert OpenShift-specific objects (like DeploymentConfigs, Routes, ImageStreams, SCC) to standard Kubernetes manifests (like Deployments, Ingress, PSA)?
- Day 2 operations: OpenShift comes with a lot of built-in features. With Talos, we need to build our own observability and ingress solutions. Did you find this operational burden too cumbersome?
- The shift away from SSH: How did your traditional sysadmin teams adapt to Talos's API-only paradigm?
Any insights, pitfalls to avoid, or tool recommendations would be hugely appreciated. Thanks!
5 Answers
If cost and ease of management are your priorities, moving towards a more vanilla Kubernetes setup makes sense. We’ve transitioned teams away from heavier platforms, and while the migration can be tough, especially translating OpenShift primitives and building out Day 2 operations, the long-term operational burden is usually lighter once everything is standardized. Talos is especially good if your team is ready to treat the cluster like an API-driven service rather than just a collection of servers.
I’d recommend Rancher as the most straightforward shift from OpenShift. It was the alternative we went with ten years ago, and I haven't looked back since. But to be fair, both platforms have evolved a lot since then.
Have you checked out Rancher/RKE2? It has a more traditional Kubernetes setup but comes with great features like a web interface and built-in observability and ingress options. It’s a solid choice if you’re looking to simplify things.
I’m with you on Rancher/RKE2! It’s a great enterprise option after OpenShift. Plus, it integrates well with other VM platforms and can handle VMs, too. I used to work somewhere that shifted to OKD and we spent so much time trying to replicate OpenShift features, so I’d definitely steer clear of that if you can.
Sounds like a plan! We’ve been using ArgoCD for both infrastructure and app deployments with OpenShift, which has made things easier for us. I’ve been considering RKE2 as well. Does it have a strong community backing it?
Have you tried setting up Kubernetes with Kubespray? It’s an open-source solution that lets you avoid vendor lock-in completely.
Talos has been a huge benefit for us. The transition from OpenShift felt quite seamless once we figured out the key differences in management, and it’s been great for simplifying our workflows.
That’s what we’re aiming for! We actually don’t use many OpenShift-specific features, so transitioning should be smoother. I find ACM really helpful for managing deployments, though. My team is a bit hesitant about the API-only model, but with practice, I think they’ll adapt. Initial setups can be challenging, but I believe it’s worth it!