Carbon-Aware Computing: Measuring and Reducing the Carbon Intensity of Software
summary
Built a cloud HPC pipeline for Vestas that autos-scales to 400,000 cores and uses ElectricityMap plus the Green Software Foundation carbon SDK to do live time-shifting and location-shifting. Time-shifting lowers emissions by 12–16% within 24 hours; location-shifting across EU sites cuts emissions by 30–90% (typical Amsterdam→Stockholm about 65%). Uses spot instances (70–90% cost savings), a dispatcher to pick region/time, artifact copy and retries, and a carbon-economy SDK to trade off price versus CO2; deployment enters production.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 1:52Green supercomputer challenge
- 8:12IT industry carbon footprint
- 8:53Energy efficiency in software
- 10:37Hardware efficiency
- 12:03Carbon awareness and time shifting
- 13:06Data center efficiency
- 15:14Optimizing compute and storage
- 17:44Spot instances for savings
- 19:32Carbon intensity basics
- 23:28Real-time carbon data
- 27:50Time and location shifting
- 29:20Carbon-aware architecture
- 33:13Measured carbon savings
- 34:39Software Carbon Intensity
- 37:30SCI formula breakdown
- 41:20Green Software Foundation tools
- 42:46Balancing carbon and cost
- 43:30Carbon Economy SDK
- 46:01User interface for balancing carbon and cost priorities
- 48:13Summary
- 49:14Call to action
related talks
55mCode that Won't Heat the Planet: Engineering Sustainable Software for a Better Future
19mGreen Code on the Cloud: Integrating Sustainability into DevOps Practices
41mGo Green: Reducing Emissions, Costs & Greenwashing
31mBridging Product and Engineering for a Greener Future: Embedding Sustainability into Software Teams
34mSustainable Software Engineering, Building Carbon-Efficient Applications
16mGreen Code: Transforming DevOps for a Sustainable Future
33mSmart Green Computing Cloud Native Operations
27mDevGreenOps