Karpenter : Your trusted Companion for Kubernetes Scaling
summary
Karpenter watches unscheduled pods and uses Kubernetes APIs to observe aggregate resource requests, then calls cloud compute APIs (for example Amazon EC2) directly to provision nodes sized to pack pods and reduce scheduling latency and cost. It exposes Provisioner/NodeClass-like CRDs to express instance families, architectures, spot versus on‑demand preferences, and TTLs; it gives preference to on‑demand when spot price is higher and automatically removes idle nodes after a timeout. Result: faster, cluster-aware scaling and fewer underutilized nodes.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:14Kubernetes scaling overview
- 2:38Traditional scaling mechanisms
- 4:01Cluster autoscaler limits
- 6:14Karpenter introduction
- 8:25Efficient node provisioning
- 10:35Karpenter’s provisioning logic
- 12:23Karpenter cost optimization
- 14:34Outcomes and advantages
- 15:59TTL and label management
- 17:31Q&A
related talks
22mKarpenter Efficient scaling of Kubernetes clusters
22mUse Karpenter to scale, optimize & upgrade K8s
19mKarpenter, the Future of Kubernetes Node Management
33mKarpenter: Best Practices and Cost Optimization Techniques
01h 01mJust-in-time Nodes for Any AWS EKS Cluster: Karpenter
27mHarness the power of Karpenter to scale, optimize & upgrade K8s
34mReduce Costs and Optimize on Your Cloud with Karpenter
19mJust in time Worker Nodes with Karpenter