Let your Kubernetes environment flourish
summary
A cluster of 240 workloads costs $3,687/month today while 99% of the time the cluster uses 61 CPU cores and 261 GB yet the sum of pod requests is 156 cores and 407 GB, so the cluster is nearly four times larger than needed. Analysis finds 131 resilience issues, links mis-sized pods to poor HPA and cluster autoscaler behavior, and produces YAML recommendations plus Jira tasks; the tool predicts OOMs, quantifies waste and carbon, and shows concrete fixes that apply automatically.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:39Efficiency in Kubernetes
- 3:46Right sizing Kubernetes
- 7:19Misprovisioning risks
- 9:26Manual right sizing
- 10:29Data to intelligence
- 12:00Right sizing with PerfectScale
- 16:42Automating fixes
- 17:24Resilience versus cost
- 18:50Risk categorization
- 19:56Multi-cluster insights
- 20:38Closing
related talks
44mFinOps at scale: Lessons from scaling Kubernetes workloads
33mThe Big Kubernetes Mistake Paying More Than Necessary
15mDemo: Resource(Cost) optimization on planet scale Kubernetes clusters.
47mOptimizing Kubernetes Resource Requests/Limits for Cost-Efficiency and Latency
33mUncover Kubernetes cost-optimization trends
29mRight-Size Your Kubernetes Deployments
39mOptimizing Resource Usage in Kubernetes
37m73,000 Pods a Day, Lessons From Misadventures In Multi-Tenant