Declarative Underlays: Scaling Purpose-Built Infrastructure(Clusters)...
Talkat KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 · 2026-09-07..09-09·
summary
Runs OpenStack on Kubernetes using Cluster API to declare and manage many purpose-built clusters, consolidating control planes into a management cluster where control plane components run as pods and use a load‑balancer on TCP/6443 and connectivity tunnels on TCP/8132. Operates across 17 data centers with per-cluster K8s versions (example: 1.31 compute vs 1.35 AI GPU), shrinks provision time from weeks to minutes, reduces a 300+ CRD mess, and enables move GPUs between inference and training to raise utilization and rebalance capacity on demand.
AI-written summary · about
chapters (12)
- 0:00Samsung SDS cloud intro
- 2:08OpenStack on Kubernetes at scale
- 4:39Single cluster challenges
- 6:44Split clusters by purpose with Cluster API
- 10:21Cluster API deep dive
- 17:36Benefits and use cases
- 21:07Scaling and upgrading clusters
- 22:10Independent tech stacks
- 23:37Optimizing cloud resources
- 25:26Infrastructure as code and AI
- 28:15Global cluster fleet management
- 29:41Conclusion and roadmap
related talks
03mFrom 5 to 1,300+ Clusters: Declarative Scaling for Private Kubernetes
12mCluster API To Streamline Kubernetes Clusters Lifecycle Management...
41mHosted Control Plane on Kubernetes (HPC)
37mManaging and Running Multiple Kubernetes Clusters in Hybrid Setups
07mBuilding Kubernetes-native applications
01h 12mOpenStack brings Kubernetes to the Edge
45mLarge-scale kubernetes deployment, which tools to use and how - Rachid Zarouali
36mManaging und Running Multiple Kubernetes Clusters