Mastering Cloud Cost Optimization with Kruize
summary
Open-source Kruize connects to Prometheus (or OpenTelemetry) inside a Kubernetes/OpenShift cluster, pulls 1/7/15‑day metrics and returns container recommendations in under five minutes (demo shows ~3.5 minutes). It reports cost vs performance choices (60th vs 98th percentile), surfaces JVM, Quarkus and Spring Boot tunables, and runs hyperparameter tuning with Optuna/TPE to run trials — one case cuts response time by 83% — and is extending GPU (DCGM) metrics.
AI-written summary · about
chapters (16)
- 0:00Introduction to cost optimization
- 1:47AI workload cost challenges
- 3:54Hyperparameter optimization basics
- 5:59Monitoring use case issues
- 6:43Cruize architecture overview
- 7:25Demo setup and installation
- 10:38Experiment configuration explained
- 12:01Recommendations in the UI
- 13:27Cost vs performance trade-offs
- 14:30OpenShift integration details
- 15:33Autotune complex objectives
- 18:22Autotune architecture explained
- 20:55HPO in non-Kubernetes contexts
- 22:01Debugging with HPO
- 23:02Future optimization areas
- 25:12Resources and community
related talks
32mOptimizing Application Performance on Kubernetes
26mDescubra las tendencias de optimización de costos en Kubernetes
33mUncover Kubernetes cost-optimization trends
57mKubernetes Cloud Cost Monitoring with OpenCost & Optimization Strategies
33mKubernetes cost optimization - How to actually do it
34mOptimizing operations & container management
26mBuilding Cost Visibility For Your Kubernetes Clusters
07mOpenCost: Cost And Resource Management