How to make Dev and Ops folks look good as an SRE
summary
Automates Kubernetes pod sizing with machine learning that consumes two weeks of historic metrics and runs rapid experiments via load tests to output deployable YAML or auto-patch deployments. Uses Stormforge Optimize with modules Live for continuous production tuning and Pro for staged load testing and per-release checks. Exposes risk levels (high/medium/low), supports Prometheus and Datadog, covers Java and other apps via env vars, and reports typical resource savings of 30–60% while maintaining or improving performance.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:22Problem with resource optimization
- 2:08Complexity in Kubernetes
- 6:22Over-provisioning vs manual tuning
- 8:28Automating resource optimization
- 10:56Load testing for peak scenarios
- 13:19Optimizing test environments
- 15:46Live optimization demo
- 17:32Results and savings
- 19:59Holistic solution and next steps
related talks
49mSite Reliability Engineering with Kubernetes
29mRight-Size Your Kubernetes Deployments
11mK8s Resources Ensuring Performance and Optimization
30mPut the ‘Auto’ in Autoscaling — Erwin Daria
11mCost and Resource Optimization In Kubernetes
30mHow Workload Rightsizing Increases Availability
30mPragmatic SRE for Kubernetes in the Cloud
25mIntelligent Kubernetes Workload Optimization