K8s Resources Ensuring Performance and Optimization
summary
Describes Scaleops, a self-hosted, in-cluster platform built on Kubernetes that uses agents and Prometheus to collect metrics, a recommender that writes CRDs, and an admission webhook that mutates pods without changing manifests. Explains an automated policy-driven workflow that classifies workloads (production vs cost) and applies per-workload/namespace/cluster optimizations, reporting a case where an EKS cluster’s vCPU usage dropped 51% (668→327) while selectively scaling up some workloads for performance.
AI-written summary · about
chapters (9)
- 0:00Pod scaling in K8s
- 1:04Resource scaling challenges
- 2:05Scaleops architecture
- 3:32Scaleops setup
- 5:09Cluster waste analysis
- 6:15Optimization policies
- 8:47Real-time tuning policies
- 9:51Auto scaler recommendations
- 10:12Optimization case study
related talks
11mCost and Resource Optimization In Kubernetes
29mRight-Size Your Kubernetes Deployments
22mHow to make Dev and Ops folks look good as an SRE
46mFinOps: Optimize for K8s
31mFailing to Autoscale?
44mFinOps at scale: Lessons from scaling Kubernetes workloads
14mOpen Source Ecosystem: Building Sustainable Solutions with Kubernetes
32mOptimizing Application Performance on Kubernetes