News
Handling the elasticity of Kubernetes
At Edgecase 2026, Constantino Vázquez, Head of Systems Engineering at OpenNebula Systems, showed how infrastructure and Kubernetes are growing ever closer together. Kubernetes is usually seen as the layer where you run and manage applications. OpenNebula turns that around and treats Kubernetes itself as an application, running on top of virtualized infrastructure.
OpenNebula is a European open-source project that has been around for about 18 years. It helps organizations build and manage private clouds. At Edgecase, the focus was mainly on Kubernetes. According to Vázquez, the biggest challenge isn’t running a cluster. It starts when multiple teams begin working with it. Clusters need to be deployed, access rights set up, and environments kept manageable. As usage grows, so does the management workload. The OpenNebula Kubernetes Service (OneKS) is built to handle this “elasticity” of Kubernetes, Vázquez explains.
Virtual manager
The advantage of this approach is that service providers no longer have to manage the Kubernetes infrastructure themselves. “At its core, OpenNebula is a virtual manager,” says Vázquez. “It can plan capacity, deploy virtual machines, isolate workloads, offer hardware profiles and automatically add worker nodes when needed.” The newest service is OneKS, which also makes it easy to create, scale, upgrade or recover clusters.
During a breakout session, OpenNebula Systems showed how new clusters can be deployed from a central interface. Behind the scenes, the platform takes care of configuring and installing the required components. For development teams, Kubernetes feels more like a service, while management and governance stay centrally organized.
Edgecase 2026: Gotta read 'em all
All the speakers, sessions, and insights from Edgecase 2026 in one place.
AI-factories
Besides Kubernetes, Vázquez also talked about AI infrastructure, a growth area for OpenNebula Systems. But that growth calls for attention to sovereignty, without losing the open-source character. “AI factories are popping up everywhere. There’s massive investment in GPUs, data centers, and networking and storage solutions. Europe needs to make sure it doesn’t fall behind the US and China.” That’s exactly where OpenNebula’s big advantage lies, he argues: it’s a European project. According to Vázquez, the software is developed and supported within the European Union and built on open APIs and standards. That said, OpenNebula Systems works closely with Nvidia, which has certified the company as a cloud provider and supporter of Nvidia technologies.
Elastic Capacity Manager
The last development Vázquez presented is GPU as a Service, a model that makes smarter use of capacity. How does that work? “A customer might pay for the guarantee that they can use eight GPUs at any time, while in practice they only need two. The remaining capacity can then be offered to other users at a lower rate,” Vázquez explains. This lets cloud providers spread capacity across different service levels. To make this possible, OpenNebula Systems developed an Elastic Capacity Manager, which it will be able to offer very soon.
