How do we choose between managed Kubernetes and self-hosted VMs?

[Joseph_J]

I'm trying to decide if we should move our container orchestration to a managed Kubernetes service or stick with our self-hosted cluster on VMs. The operational overhead is becoming a real pain, but I'm worried about losing fine-grained control over the underlying nodes and networking stack.

[Edward.J]

We switched to a managed Kubernetes service a few months ago and the ops grind is finally tolerable. Patching, upgrades, and node health checks used to eat a lot of time; now those run mostly unattended. We still care about networking policy and node sizing, but the day to day is lighter.

[Emily64]

Not sure the problem is the control surface. Our dashboards and alerting felt like a moving goalpost; we trimmed alerts, clarified ownership, and still end up firefighting on nights or weekends.

[SamuelYA]

Are we sure the real blocker is the orchestration layer, or is the bottleneck elsewhere like CI pipelines and image builds? I keep wondering if the cluster is the symptom rather than the cause.

[Luke76]

During planning we tracked deployment time, and after moving away from self-hosted, our deploys completed faster but we hit a new constraint around multi-region networking and service discovery. Not sure if it's worth the lock-in tradeoff yet.