Art of decluttering your Kubernetes cluster
summary
Clusters accumulate forgotten CRDs, Helm releases and idle nodes that inflate cost and risk; an audit finds 270 CRDs and dozens of unused Helm charts. Practical fixes include running simple kubectl queries, using kubectl delete for stale objects, labeling and namespace conventions to attribute spend, setting end dates and autoscale limits for ephemeral nodes, and automating reclamation with Cube capacity and Cube Center. Expect measurable CPU/RAM reductions and smaller instance types (eg xlarge → large).
AI-written summary · about
chapters (9)
- 0:00Decluttering Kubernetes intro
- 1:22Why clusters need decluttering
- 3:51Consequences of clutter
- 4:35Identifying cluster clutter
- 7:29Monitoring cluster efficiency
- 8:33Practical decluttering steps
- 10:40Optimizing resources and costs
- 11:22Namespace and label strategies
- 12:04Q&A
related talks
14mCut Costs, Not Performance: Implement Kubernetes Cost Optimization Strategies
29mRight-Size Your Kubernetes Deployments
36mOptimizing Resource Usage in Kubernetes
47mPostmortem: A Modern Approach to Managing Kubernetes Clusters
28mScaling in Kubernetes Safely on On-Prem KaaS Across 1,300+ Clusters and 40,000+ Nodes
34mBecoming an Expert of Diagnosing a Kubernetes Control Plane Problem
36mKubernetes Confessions: Real Mistakes We’ve All Made (and What They Teach Us)
44mFinOps at scale: Lessons from scaling Kubernetes workloads