Does Green Software need Open Hardware?
summary
Open Compute Project shows concrete work: an open telemetry-aware scheduler, a GPU-aware scheduler and an open power manager (next release adds P-state/C-state control) that together aim to cut server power use by ~30–40% in trials. Operators run OCP kit in production and a distributed orchestrator called Krake moves workloads between sites for price, heat reuse or weather. Liquid immersion cooling (water ~3,500x air heat capacity) and DC power shelves are tested; industry data cites 10 MW racks ≈ power for 20,000 households and Bitcoin energy exceeding Switzerland (2019).
AI-written summary · about
chapters (13)
- 0:00Introduction to OCP and panel
- 4:44Panelist introductions
- 7:32Defining green software
- 12:16Data center infrastructure
- 15:00Sustainability and performance
- 18:11Importance of open hardware
- 20:19Efficiency in facilities
- 23:22Software scheduling and power
- 26:44Hardware longevity and metrics
- 31:16Data collection and visibility
- 34:24Future of sustainable design
- 39:29Bridging the facility-IT gap
- 43:24Closing thoughts and Q&A
related talks
21mIntro to the Open Compute Project OCP
56mA Developer's Perspective On Greener Future: GreenOps in practice
27mGreenOps Sustainable engineering 1
36mOpenPOWER and GPU Porting Platform for FOSS Using OpenStack at OSUOSL
06mPOWER your Cloud
27mSustainable, Modern, and Flexible Solutions for Cloud Environments
24mWhy Data Centers Use OSS and How They Do It
26mGoing Green: The Future of Sustainable Software