vCluster Labs (formerly LoftLabs), the company pioneering Kubernetes virtualization, is introducing the latest capability for vCluster, Auto Nodes—enabling node autoscaling for virtual clusters across any environment, including public cloud, private cloud, hybrid infrastructure, and even bare metal, using Karpenter, the open source autoscaler created by Amazon Web Services (AWS).
Auto Nodes is infrastructure-agnostic and extensible by design, according to the company. Platform teams can use Terraform or OpenTofu to declaratively define infrastructure and even build custom providers for any non-standard environments.
Besides Terraform and OpenTofu, vCluster’s Auto Nodes also ships with native support for tools such as NVIDIA Base Command Manager (BCM) and KubeVirt which make it easy to autoscale workloads across bare metal private clouds and AI clusters with NVIDIA GPUs.
“With Auto Nodes, we’re unlocking Karpenter-powered dynamic autoscaling across any infrastructure,” said Lukas Gentele, CEO of vCluster. “It’s not just about autoscaling, it’s about doing it anywhere, with full isolation and flexibility—all based on a proven open source technology. That’s a superpower no other K8s distro has across private cloud and bare metal environments.”
vCluster recently introduced Private Nodes, allowing users to join dedicated Kubernetes nodes directly into a virtual cluster. This made it possible to run workloads in fully isolated, single-tenant environments while retaining the lightweight benefits of vCluster’s containerized control plane.
Now with Auto Nodes, isolation no longer comes at the risk of losing scalability or increasing infrastructure waste. Each virtual cluster can now scale its Private Nodes automatically, on any cloud, in any environment, using a built-in Karpenter integration, the company said.
Auto Nodes builds on the Private Nodes tenancy model by embedding a Karpenter operator directly into each virtual cluster. The Karpenter operator:
- Monitors pods (including unschedulable ones) inside the virtual cluster;
- Dynamically provisions new nodes that match specific constraints;
- Automatically removes unused nodes again once workloads terminate and load reduces;
- Works with a range of node providers, including Terraform, BCM, KubeVirt, and custom environments; and,
- Supports GPU and CPU-based workloads across diverse hardware types.
With Karpenter’s just-in-time autoscaling and vCluster’s multi-provider support, each virtual cluster becomes its own fully isolated autoscaling unit, able to request exactly the compute it needs, when it needs it, from whatever infrastructure is available.
This unlocks hybrid and multi-cloud architectures where virtual clusters can elastically scale GPU or CPU workloads without sacrificing isolation or duplicating clusters, according to the vendor.
Auto Nodes makes it possible to shift workloads between cloud providers or data centers dynamically, enabling cross-cloud cost optimization based on real-time pricing, availability, or policy constraints, without changing application code or infrastructure setup.
Auto Nodes is now available as part of the vCluster v0.28 and vCluster Platform v4.4 release. It supports static and dynamic node pools, fine-grained constraints via NodeTypes, and can target specific instance shapes, GPUs, and architectures.
For more information about this news, visit www.vcluster.com.