Green Software Architecture - Dos Don'ts and Some Surprises
summary
IT currently uses about 10% of global electricity and ~2–3% of CO2, forecast to rise, so saving power matters; biggest, cheapest wins come from operations: switch off unused test environments, right‑size cloud instances and use autoscaling/multi‑tenancy to raise server utilization (idle servers still draw ~60% of peak). Developers lack real‑time power APIs, so use proxies like CPU/GPU cycles, optimize DB indexes (seconds→ms), prefer compact binary protocols and consider ML costs (training a big model can equal the lifetime emissions of ~five cars).
AI-written summary · about
chapters (10)
- 0:00Green software intro
- 3:04Why green software matters
- 7:26Energy-efficient challenges
- 17:45Requirements for green software
- 28:51Engineering energy efficiency
- 38:50Design choices for carbon
- 50:57Languages and green software
- 1:04:37Testing limits and wins
- 1:13:07ML’s environmental cost
- 1:17:45Future tools and practices
related talks
46mMaking Software & Data Architectures Sustainable
38mArchitecting for a Greener Future
30mFighting Climate Change by Building Sustainable Software
44mWhat Is Green Software?
42mEnergy-Efficient Software Architecture for Developers
43mGreen Software – How To Make Our Applications Carbon Aware?
49mHow to Save the Planet Without Leaving your Desk!
39mCreate an Eco-Friendly World with Green Software Engineering