Optimizing Kubernetes Resource Requests/Limits for Cost-Efficiency and Latency
summary
Collected metrics from 80+ clusters reveal a large fraction of pods run without proper memory or CPU requests, causing scheduling slack, unexpected OOM kills when limits are too tight, and wasted cloud spend. A practical pipeline uses historical pod-level CPU/memory samples and percentile-based sizing to set requests and conservative limits, tested with canary rollouts across hundreds of services. All components are open source tools, and the outcome is reduced cost and lower latency impact for latency‑critical workloads.
AI-written summary · about
chapters (7)
- 0:00Introduction to resource management
- 7:27Understanding requests and limits
- 16:06Quality of Service in Kubernetes
- 25:19Step-by-step optimization approach
- 33:24Cluster scaling strategies
- 40:27Cost and performance trade-offs
- 46:15Summary and key takeaways
related talks
29mRight-Size Your Kubernetes Deployments
39mStaying in Tune: Optimize Kubernetes for Stability and Utilization
33mUncover Kubernetes cost-optimization trends
15mDemo: Resource(Cost) optimization on planet scale Kubernetes clusters.
21mLet your Kubernetes environment flourish
33mThe Big Kubernetes Mistake Paying More Than Necessary
44mFinOps at scale: Lessons from scaling Kubernetes workloads
38mA Tale of Tail Latency: Understanding k18n CPU Requests and Limits for Sust. and Profit