What if your Kubernetes clusters could scale all the way down to zero? We’ve rolled out a powerful update to the Magnum Cluster API driver that gives you more control, more efficiency, and a whole new level of flexibility.
We’re introducing a powerful new update to the Magnum Cluster API driver from VEXXHOST. With this new release, you can now deploy Kubernetes clusters with node_count set to zero - that means only the control plane is created by default. From there, our built-in autoscaler does the rest. When min_node_count is set to zero, your node groups can now scale all the way down to zero, giving you more flexibility and even better cost efficiency in your OpenStack environments.
Zero-Worker Clusters on Magnum
Until now, a default node group was automatically created during cluster deployment. Now, with Magnum, you simply create your cluster with node_count set to zero. Here’s what that unlocks:
- Control Plane-Only Deployment - Only the control plane is deployed initially, giving you complete control over when and how worker nodes are added.
- Custom Node Group Management - Build your cluster from the ground up with custom node groups tailored to your specific workload requirements.
- Optimized Resource Utilization - Start with zero worker nodes, and scale out only when the demand hits, saving both compute and budget.
Auto Scaling Down to Zero
Alongside zero-worker clusters, we’ve made our autoscaler even smarter. Now, when you set min_node_count to zero, your node groups can scale all the way down to zero and back up again as traffic or workload demands rise.
It uses the same autoscaling mechanism that’s already baked into the driver. There’s no extra setup required. Just set the min_node_count to zero and let the autoscaler handle the rest.
Why It’s a Big Deal
Whether you’re working with short-lived workloads, batch jobs, or just want to avoid burning resources during off-hours, scale-to-zero makes your Kubernetes clusters work smarter. You can start with just the control plane and build your infrastructure exactly as needed with custom node groups.
The system reacts to real-time demand.
So, you get dynamic resource allocation that scales down when it should, scales up when it must, and keeps your costs in check the entire time. This way you get to maintain full control over node provisioning without sacrificing performance or responsiveness.
It’s a cleaner, more cost-efficient approach to running Kubernetes on OpenStack. And it gives you a lot more breathing room when it comes to cluster design and scaling strategy.
Continuing to Push Kubernetes Forward
As more organizations adopt Kubernetes for dynamic, event-driven workloads, the ability to start small and scale intelligently becomes critical. By enabling control plane-only clusters and true scale-to-zero autoscaling, Atmosphere is helping teams move toward more responsive, resource-efficient infrastructure without compromising on flexibility or performance.
At VEXXHOST, we’ve been building cloud management infrastructure for nearly two decades. So, this new feature is part of a larger effort to build tight integrations that simplify operations, reduce overhead, and give users more ways to shape infrastructure to their needs.
Get Started
There’s more to come. But for now, if you're exploring how to fine-tune your Kubernetes clusters, or want a cloud that responds to your workloads instead of the other way around, we’re here to help. Reach out to our team or explore the documentation for a closer look at how it works.
