Cluster API To Streamline Kubernetes Clusters Lifecycle Management...
summary
NVIDIA consolidates on-prem Kubernetes lifecycle by building a custom ClusterAPI infrastructure provider for bare metal that supports heterogeneous nodes (CPU, GPU, TEGRA, ARM). Cluster definitions become declarative YAML, templates are wrapped in Helm charts so upgrades are a single values.yaml change, and Flux CD GitOps applies and reconciles changes. Result: automated provisioning, reduced manual toil, and operational scaling to roughly 75 clusters under a single management cluster.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:44Siloed Kubernetes operations
- 2:58Cluster API and adoption
- 4:21Custom infra provider on-prem
- 5:47Cluster API with Helm
- 8:13GitOps with Flux CD
- 9:23Key takeaways
- 11:12Q&A
related talks
25mCluster API: IaC for Kubernetes
31mDeclarative Underlays: Scaling Purpose-Built Infrastructure(Clusters)...
32mDon't Start With 500 Clusters: Patterns in Scaling Multi-Cluster Kubernetes…
31mHow to Migrate 700 Kubernetes Clusters to Cluster API with Zero D...
46mSIG Cluster Lifecycle (Cluster API)
25mIntroduction To Cluster API
06mSponsored Keynote: Recursive Kubernetes: Cluster API and Clusters as Cattle
06mAutoscaling GPU Clusters Anywhere — Hyperscalers, Neoclouds & Ba...