Uploaded August 2026 | Updated September 2026, 2 weeks ago
Don't miss out! Join us at our next KubeCon + CloudNativeCon events in Shanghai, China (8-9 September, 2026) and Salt Lake City, United States (Nov 9–12, 2026). Connect with our current graduated, incubating, and sandbox projects as the community gathers to further the education and advancement of cloud native computing. Learn more at kubecon.io
Towards Sustainable Multi-Cluster Management in Real-World Academic Infrastructure - Jinwang Mok, GIST NetAI Lab. & ChangHyeon Im, GIST | Gwangju Institute of Science and Technology
Research labs face distinct challenges: heterogeneous hardware, stateful services, and student operators who learned Kubernetes on the job—accumulating legacy no one dares untangle.
Our environment: four bare-metal clusters over 100GbE—a 2PB Rook-Ceph lake, a 22-GPU cluster (7 types incl. L40S, A100), edge devices—run by grad students. We call our unified multi-cluster target "ScaleX-POD". Migrating demands two hurdles: risk-free live migration and tenant isolation.
First, extracting a management cluster from production while external tenants actively run—transferring ArgoCD ownership without orphaning pods or storage, dry-running via vcluster. We share what broke and what we would change. Second, Kyverno-enforced isolation: PSS Restricted, default-deny NetworkPolicies, namespace-scoped RBAC—containing blast radius during and after migration.
Attendees gain failure lessons, migration trade-offs, and a PR-based workflow to reduce knowledge loss across operator transitions.
Don't miss out! Join us at our next KubeCon + CloudNativeCon events in Shanghai, China (8-9 September, 2026) and Salt Lake City, United States (Nov 9–12, 2026). Connect with our current graduated, incubating, and sandbox projects as the community gathers to further the education and advancement of cloud native computing. Learn more at kubecon.io
Towards Sustainable Multi-Cluster Management in Real-World Academic Infrastructure - Jinwang Mok, GIST NetAI Lab. & ChangHyeon Im, GIST | Gwangju Institute of Science and Technology
Research labs face distinct challenges: heterogeneous hardware, stateful services, and student operators who learned Kubernetes on the job—accumulating legacy no one dares untangle.
Our environment: four bare-metal clusters over 100GbE—a 2PB Rook-Ceph lake, a 22-GPU cluster (7 types incl. L40S, A100), edge devices—run by grad students. We call our unified multi-cluster target "ScaleX-POD". Migrating demands two hurdles: risk-free live migration and tenant isolation.
First, extracting a management cluster from production while external tenants actively run—transferring ArgoCD ownership without orphaning pods or storage, dry-running via vcluster. We share what broke and what we would change. Second, Kyverno-enforced isolation: PSS Restricted, default-deny NetworkPolicies, namespace-scoped RBAC—containing blast radius during and after migration.
Attendees gain failure lessons, migration trade-offs, and a PR-based workflow to reduce knowledge loss across operator transitions.
