Authors: Hannah Smith and Ismael Velasco.
Reviewers and contributors: Michael J. Oghia, Fershad Irani, Wim Vanderbauwhed.
Additional informal input and advice: Phillip Jenner and Chris Adams.
-
The core concept of shifting compute jobs to respond to the grid is a smart one. But there are only small-to-zero carbon reduction benefits from implementing most current carbon-aware time-shifting and location-shifting computing approaches.
-
If such approaches are adopted at scale without implementation constraints, they are likely to increase emissions and destabilise electricity grid systems. This does not improve the tech industry’s contributions to global sustainability, but rather worsens them. It also risks becoming, or is already, a greenwashing effort.
-
The key failure is never applying appropriate warning labels to draw attention to key caveats. Such as
-
Carbon-aware computing that can yield actual carbon reductions runs when demand is low using curtailed electricity in stable grids, or on genuinely additive renewable electricity.
-
“Is your compute’s net electricity demand reducing?” is the central question if we really want to make computing more sustainable. Acknowledging Jevons paradox and accepting that optimisation alone isn't enough, means we must find ways to limit the growth of computing's resource use.
- We need to usher in a more mature, holistic and nuanced approach to reducing the CO2 emissions from running software. For now we’re calling this ‘grid-aware computing’.
This work is open source and licensed using Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0 Deed).
In short you are free to use this work, but you must give appropriate credit and link back to this repo.
https://github.com/climateaction-tech/grid-aware-software.
This body of work is divided into seven distinct sections:
- Houston, we have a problem
- What software engineers need to know about how the grid works
- What’s the problem with carbon-aware software then?
- When does carbon-aware software make sense?
- Addressing the elephant in the room
- Where do we take carbon-aware from here? Introducing grid-aware computing
- What can you do to help?
Appendices
The body of work presented here, started after Ismael posted some questions and research into the CAT's Slack. Hannah, amongst many others picked up on the discussion.
Following some debate on the issue, Hannah and Ismael agreed to work together to further develop the research on the problems with carbon-aware and ultimately come up with this presentation of the issues.
If it were not for CAT this work would not have come to fruition on this way - thank you CAT for bringing people together!