Architecting for Sustainability: A Software Architects Guide to Green Software Keynote
summary
Software contributes 2–4% of global emissions and data centers use about 1.5% of global electricity in 2024 (forecast ~4.5% by 2030). A machine learning batch (≈10 minutes) running tens–thousands/day measures ~125 gCO2e in one region; moving region saves ~25% and shifting time saves ~15%. The team builds an Azure pipeline that queries the electricityMaps API, wraps jobs with the green-metrics tool and implements automatic demand shifting with a manual bypass, cutting estimated emissions by nearly half in practice.
AI-written summary · about
chapters (12)
- 0:00Software’s emissions matter
- 4:02Who is Owen Woods and his background
- 6:08Emissions in software systems
- 12:32Green Software Foundation principles
- 16:21Demand shifting to reduce emissions
- 18:10Demand shaping for lower-carbon computing
- 20:15Moving workloads to client devices
- 22:46Measuring software emissions
- 25:53Sustainability challenges in AI
- 32:37Case study: Demand shifting in ML
- 39:50Practical steps to start
- 47:32Resources for green software
related talks
41mWriting Greener Software Even When You Are Stuck On-Prem
26mGoing Green: The Future of Sustainable Software
34mSustainable Software Engineering, Building Carbon-Efficient Applications
36mArchitecting in the Cloud for Sustainability
56mA Developer's Perspective On Greener Future: GreenOps in practice
01h 00mSustainable Azure - Tools & Patterns for Carbon Aware Cloud Applications
28mHow to limit your cloud impact, from bare metal to AI
34mEmbracing Sustainability in Cloud-Native Ecosystems