Cluster API & Cluster Autoscaler
summary
Combines Cluster API (CAPI) CRDs—Cluster, MachineDeployment, MachineSet, Machine—and provider implementations (Magalu Cloud/OpenStack/VM-based) with Cluster Autoscaler to automate on‑prem Kubernetes lifecycle and node autoscaling. Describes achieving reproducible clusters up to 254 worker nodes with typical provisioning in 3–5 minutes (VM creation ≈102s), using Autoscaler to scale up for Pending pods and scale down for FinOps savings, and contrasts provider tradeoffs (Carpenter/AWS tighter integration versus CAPI’s extensibility).
AI-written summary · about
chapters (11)
- 0:00Introduction
- 4:21What is Cluster API
- 8:25Cluster API components
- 13:13Technical advantages
- 17:13Fast cluster provisioning
- 21:43Cluster API + Autoscaler
- 26:26Practical benefits and use cases
- 27:55Infrastructure contribution
- 30:09Magalu Cloud future
- 36:50Disaster recovery
- 38:16Closing
related talks
25mIntroduction To Cluster API
32mRun Cluster Autoscaler on Your Own Cloud; or, Why Do We Need Cluster API?
32mDon't Start With 500 Clusters: Patterns in Scaling Multi-Cluster Kubernetes…
36mCluster Autoscaler Evolution
32mBeyond CAS: Why the world needs another Kubernetes Cluster Autoscaler
21mWhy should you bother about cluster autoscaling- Karpenter
27mCluster-as-a-Service
32mAutoscale Your on-Premises Bare Metal Kubernetes Clusters
