Kubermatic branding element

Scaling Kube-Apiserver to Thousands of Tenants: Lessons Learned from kcp

Video

Watch Simon Bein's talk

kcp has become a powerful solution for multi-tenant control planes. Abstraction is only half the battle, though, and the most common question from previous talks has been: “How does it actually scale?” In this session, Simon Bein shares results from the kcp project’s latest load-testing efforts on 10,000 logical clusters and derives best practices for scaling a kcp instance. The talk examines the internals of the kube-apiserver and shows how it behaves under the pressure of massive multi-tenancy. Whether you already run kcp or are simply curious about the apiserver’s limits, this session provides a blueprint for managing control planes at high scale.

Key Takeaways

  • Real Load-Testing Data: What the kcp project learned from testing at 10,000 logical clusters.
  • Inside the Kube-Apiserver: How the apiserver behaves under massive multi-tenancy and where its limits appear.
  • Scaling Best Practices: Practical guidance for sizing, tuning, and operating a kcp instance at scale.

Leading Companies Choose Kubermatic

Allianz Vonage CNCF Cube FHE3 Switch inventx Datagroup Krone Charite Justus-Liebig-Universität Gießen Heidelberg University Swisscom Datev BWI