Put the ‘Auto’ in Autoscaling — Erwin Daria
summary
Storm Forge deploys optimize live as a lightweight agent single pod that scrapes cAdvisor metrics, sends them to a SaaS ML backend, and emits YAML patch suggestions (deployment + HPA) that can be auto-applied on a schedule or exported into GitOps. It claims initial recommendations in 60–90 minutes, stable weekly profiles in ~7 days, finds ~69% CPU unutilized and ~18% memory, estimates $10M/year wasted at >1k nodes, reports up to ~75% cost reduction and saves ~2+ FTEs for 500 workloads.
AI-written summary · about
chapters (5)
- 0:00Introduction and speaker intro
- 3:49Workload resource autoscaling challenges
- 11:11Optimized Live demo overview
- 21:01Cluster-level optimization benefits
- 25:36Q&A session
related talks
31mFailing to Autoscale?
22mIntelligent Autoscaling in Enterprise...
22mHow to make Dev and Ops folks look good as an SRE
31mAccelerate AI adoption: automated cloud efficiency
14mA Tale of Autoscale: Increasing Resiliency while Reducing On-Call Toil
35mAutoscaling Kubernetes Deployments: A (Mostly) Practical Guide
28mScale Smarter Not Harder: How Extending Cluster Autoscaler Saves Millions
09mAI-Powered Cloud Automation – transforming cloud operations