More Data Please: Hands on Green Cloud Experiments
Talkat KubeCon + CloudNativeCon Europe 2025 · on Wednesday, 2025-04-02· at 15:15· Level 0 | ICC Capital Hall | Room J·
summary
A four-node bare‑metal setup runs a k3s cluster (Cilium) on Intel mini‑PCs and measures energy with both a plug meter and Scafandra software, revealing plug readings near ~40 W while Scafandra reports under 5 W until an idle hardware offset is applied. The workflow captures Day 0 to Day 2 steps: manual k3s deploy, ArgoCD and Falco added, then Prometheus/Grafana/OpenTelemetry and Kepler for observability, while stress-ng load + Falco monitoring and experiments on language/OS/VPA/KEDA show which changes actually move watts.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:16Measure energy at hardware
- 5:00Bare-metal test environment
- 7:31K3S cluster with NixOS
- 10:26Energy measurement methods
- 13:13Scope of energy measurements
- 14:41Experiment design
- 16:51Power in idle cluster
- 18:53Workloads impact on power
- 20:43Analyzing energy data
- 22:31Future experiments
- 25:42Q&A
related talks
33mSmart Green Computing Cloud Native Operations
20mMonitoring Energy Consumption and Building Energy Efficient Systems the Cloud Native Way
35mSustainability Research the Cloud Native Way
26mAccelerating Science, Decelerating Carbon: Sustainable Computing...
28mExploring Kepler's Next Chapter: Achieving Cloud Native Sustainability With MCP Integration
18mKeynote: Sustainability Chronicles: Innovate Through Green Technology with...
25mScaphander hands on - demo & new features
20mEnergy Efficient Placement of Edge Workloads