Kepler Watt Store: Kepler Software Watt Watcher Store
summary
Kepler Watt Store packages Kepler and Peaks into concrete tools: Kepler (eBPF → Prometheus) exports per-process/container/pod power using RAPL, NVML, ACPI, Redfish/IPMI or regression-trained models and runs on bare metal and VMs; Peaks (ML-based scheduler) uses pre-trained models mapping node utilization to power to pack workloads and report lower cluster power than Kubernetes’ default spread; prototypes include CI GitHub Actions and a Quarkus extension for local dev power profiling; Kepler is a CNCF Sandbox project and IBM/Intel have used it for architecture comparisons.
AI-written summary · about
chapters (12)
- 0:00Kepler Watt Store intro
- 2:58Software power consumption
- 7:24Kepler’s cloud sustainability
- 11:40Kepler Watt Store tools
- 14:10Power monitoring in dev
- 16:19CI energy checks
- 17:03Runtime monitoring with Kepler
- 17:44Peaks power-efficient scheduling
- 19:32Sustainable software approach
- 20:35Q&A language integration
- 22:47Real efficiency improvements
- 24:12Emissions-aware scheduling roadmap
related talks
20mKepler - Optimized power usage monitoring for Kubernetes
29mPower Efficiency Aware Kubernetes Scheduler
27mGreening the AI Cloud: Validating Power Models for Kubernetes Containers
23mFrom Pods To PUE: A Digital Twin Approach To Sustainable Kubernetes Infrastructure
30mSustainability Chronicles: Innovate Through Green Technology
12mKeynote: Cloud Native Sustainability - How You Can Help Right Now!
15mOptimizing Application Performance on Kubernetes
29mEdge DC Energy Efficiency: A K8s Workload Allocation Optimizer with Generic Server Power Model