By Matthias Galster, University of Calgary, Canada
Abstract: Life-cycle assessment, as applied to conventional product development, can be applied in the Software Engineering domain to determine the environmental impact of a software product itself, but also the environmental impact of a software development process.
Advertisement
5 comments
Comments feed for this article
April 19, 2010 at 11:09 am
anonymous
Summary: The paper proposes the use of Life-Cycle Assessment (LCA) as a scaffolding for determining environmental impact.
Writing: well structured and clearly expressed
Suggestion: the paper could benefit from a real example. Without it, the author’s claim is uncertain.
Specific issue: does the Triangle Tool support more than three alternatives? If not, it looks like what is needed is some form of linear programming.
General concern: any time weights are used, there is a danger of arbitrariness; that is, the conclusions are only as good as the choice of weights.
Overall recommendation: if an example is included, then worthy of inclusion but not worthy of an extended presentation.
April 22, 2010 at 2:48 pm
Ian Sommerville
I think the general notion of life-cycle assessment is sound but common sense suggests that applying this to software engineering activities isn’t going to lead to much energy saving. We really have to focus on the big users of energy and not be introspective and simply look at ourselves.
Software engineering is hard enough and the practical reality is that energy considerations in software are always going to be so marginal that no-one would risk adding this to their development processes.
April 23, 2010 at 10:47 am
Steffen Zschaler
It seems reasonably accepted that energy balances should ideally be based on life-cycle assessment decisions. This clearly also applies for software-based products. So, this paper is well within the remit of this workshop. However, there are two areas where I feel the paper doesn’t go far enough:
1. The key issue applying LCA to software is of course determining the ecologigal or environmental impact of software. The rest is relatively straight-forward multi-criteria analysis. So, this should be the research area to be addressed.
2. Surely the energy footprint of software development processes or the software itself is not going to be sufficiently substantial to allow for huge improvements. However, a true life-cycle analysis would require to understand the effects software has on its environment and how it potentially changes energy consumption of other elements in its environment.
April 28, 2010 at 5:54 pm
Ideas for the workshop program « Software Research and Climate Change
[...] much scope is there to make software engineering itself greener? (inspired by Matthias Galster’s paper on life-cycle assessments for SE, and Jonas Helming et al’s paper on requirements modeling [...]
May 2, 2010 at 4:54 pm
Sebastian Gonzalez
I agree with Ian that focusing on big users of energy is a more pressing issue, but I disagree that software engineering is difficult enough to include energy considerations for only marginal gains. For this same reason energy considerations might not be going to make it into many industrial processes. We should avoid falling in this trap. We cannot dismiss the possibility of including LCA in SE processes without even trying it and corroborating its effects. I pretty much like this idea, including as the author suggests factors which are not usually considered in traditional SE: packaging and distribution, support services such as call centers vs field technicians, and architectural choices such as using cloud computing facilities vs using dedicated servers. What is the environmental impact of such choices? To my knowledge we have not got the faintest (measurable) idea. BTW, this all feels pretty related to the field of Industrial Ecology, wich could be referred to in the paper.