Most Kubernetes clusters run somewhere in the middle, on ordinary hardware and at an ordinary scale. The extremes are where it gets interesting, and where it becomes clear how much the parts of Kubernetes depend on each other.
These extremes are the subject of my demo at KubeCon in Salt Lake City this November. I am taking Kubernetes to both extremes with two clusters, one very small and one very large.
What I will show
The first cluster is a live demo of the world’s largest Kubernetes cluster. I’ll demo what it takes to scale Kubernetes components and workloads and how realistic it is to operate. We’ll break down which components are likely to break first and how to add resiliency.
The large cluster runs on Talos Linux and uses KubeSpan, the Talos feature that connects every node through an encrypted WireGuard mesh.
The second is the world’s smallest Kubernetes cluster that fits in the palm of your hand. We’ll review how to scale Kubernetes down and what to watch for along the way.
Neither cluster is built for production. Both are worth seeing in person, and anyone interested in Kubernetes is welcome to attend.
Where to find me at KubeCon
- Event: KubeCon + CloudNativeCon North America 2026, Salt Lake City
- Date and time: Wednesday, November 11, 1:05 to 1:25 PM
- Session: The World's Smallest and Largest Kubernetes Clusters
- Location: Demo Theater, Salt Palace, Level 1, Solutions Showcase, Hall 1, Alpine Overlook Zone
- Session page: Demo Theater: The World's Smallest and Largest Kubernetes Clusters
And at KubeAuto
Prior to KubeCon, I am also speaking at KubeAuto Day North America, a free one-day event co-located with KubeCon + CloudNativeCon North America 2026. Built for Kubernetes practitioners, the event series covers FinOps, platform engineering and Kubernetes automation, with talks drawn from production experience.
- Event: KubeAuto Day North America
- Date and time: Monday, November 9, 4:45 to 5:05 PM
- Session: The Ironies of AI
- Description: The tools we build to take work off our hands often leave us with only the hardest parts. Taken out of the day-to-day loop, we lose the context we need to understand how a system works and how to change it. Researchers raised these concerns about automation back in 1983, and AI has only amplified them. By learning from decades of automation, we can better judge when and where AI should be applied.
- Location: Clark Planetarium, 110 South 400 West, Salt Lake City
- Registration: Free, separate from KubeCon: kubeauto.day/north-america
Timea Macarie-Papp
Amarachi Iheanacho
Steve Francis