Communications of the ACM
Editor's letter
Owning Computing's Environmental Impact
For decades, we have carried the conceit that "computing makes everything more efficient," so the impact of computing on the environment is a net positive. But computing's unprecedented success has produced an explosion in its use, quantity, and direct environmental impact. It is time for the computing community to face up to computing's growing environmental impact—and take responsibility for it! And further, to undertake research, design, and operations to reduce this growing impact.a
The transition to SSDs and the consolidation of enterprise computing into efficient cloud datacenters has for a decade blunted the impact of growing computing use. But cloud computing's extraordinary scale (200TWh, $200B in 2017) and ICT's projected power growth (to 21% of global power consumption by 2030)5 drive the rapid growth of the cloud's atmospheric carbon emissions.3,6 Computing is the fastest-growing use of electric power in the developed world, and is driving the buildout of power generation and transmission in much of the developing world. If the world is to meet the Paris Accords goals for greenhouse-gas-emissions, computing must reduce its direct emissions.
Equally daunting is the rapid growth of waste from computing electronics, notably consumer products, smartphones, and the plethora of "smart devices" collectively termed the "Internet of Things." In 2016, e-waste reached 44.7 million metric tons per year, comparable to the size of the nine Pyramids at Giza, or 1.23 million 18-wheel trucks full of trash. This is an 8% increase from only two years earlier.7 Of this massive quantity, only a fraction is collected and recycled, with the largest fraction simply dumped into landfills or incinerated. Any claims that computing is "good" for the environment, must reckon with this waste problem.
Some computing professionals believe that Moore's Law or Dennard scaling mitigates these problems. Far from it, they actually exacerbate it! Efficiency is not a solution, as 19th-century British Economist William Stanley Jevons noted in 1865, "efficiency increases consumption," a rule widely known as Jevon's Paradox.4
Recycling programs are constructive, but less than 20% of e-waste is recycled.
Carbon offsets are constructive, but not enough. As regions undertake ambitious 100% renewable fraction goals—San Diego (2035), California (2045), European Union (entire economy 2045)—offsets are of decreasing benefit. Real solutions must achieve direct matching and supply following.2,8 Recycling programs are constructive, but less than 20% of e-waste is recycled—it is just not economic. Innovative approaches to capture or render benign e-waste are a critical need.
Computing technologies and systems must be designed and shaped for lower carbon and environmental impact. Here is a call to action to the computing community: Let's adopt goals equally ambitious to those of the climate community.
Let's create technologies and systems that in their manufacture, construction, and operation approach the goal of 100% carbon-free and neutral environmental impact!
Andrew A. Chien, EDITOR-IN-CHIEF
References
1. Dreyfuss, E. How Google keeps its power hungry operations carbon neutral. Wired (Dec. 1, 2018).
2. Google White Paper. Moving toward 24x7 Carbon-Free Energy at Google Data Centers: Progress and Insights.
3. Greenpeace. Clicking Clean: Who is winning the race to build a green Internet?
4. Jevon, W. The Coal Question. Macmillan, 1865.
5. Jones, N. How to stop data centres from gobbling up the world's electricity. Nature (Sept. 12, 2018).
6. Shehabi, A. et al., United States Data Center Energy Usage Report. LBNL, June 2016.
7. United Nations University. The Global E-waste Monitor 2017 (Dec. 2017).
8. Yang, F. and Chien, A.A. ZCCloud: Exploring Wasted Green Power for High-Performance Computing, IPDPS, May 2016.
Footnotes
a. Of course, I do not mean to imply that there have been no efforts to date—much to the contrary. But, rather to call for renewed and universal engagement on this agenda.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2019 ACM, Inc.
Comments
Ole Agesen
Dear Dr. Chien:
We thank you for your insightful, timely, and clear description of Computing's role in the problem of utmost importance for the 21st century: global warming and sustainability.
The holy grail of energy proportional computing has not been achieved and may never be.
The good news is that software techniques, chiefly virtualization, but also containers, serverless, software-defined networking and storage, have already had a major impact on improving the utilization of data center resources, whether owned and operated by enterprises or cloud providers. By allowing all components of a data center to be configured and monitored via software (i.e., the software-defined datacenter, or SDDC), it has been possible to drive utilization of all these resources much higher, reducing both capex and opex. The aggregate, global result is substantially reduced IT, data center and power plant infrastructure and operations and the associated avoided energy, water, carbon, land use, pollution and waste impacts.
The quantitative impact of the above is staggering. A recent IDC study [ref1] estimates that VMware's virtualization business (e.g., server consolidation) over the past 15 years has resulted in the avoidance of nearly 539 million metric tons of CO2, the equivalent of having eliminated over 1 trillion automobile miles. In 2017 alone the CO2 avoidance is the equivalent of removing over 18 million cars from the road. Followers in the industry have probably added an equivalent amount of savings.
Long term, we hold out hope that innovation may continue to deliver equally dramatic savings over the next 20 years and beyond. It may be a stretch goal, but not impossible(*). We believe this task to be an excellent business opportunity, one that VMware can help address, and one that we hope can attract the best talent to help us do so.
Sincerely,
Nicola Peill-Moelter, Nicola Acutt, Ravi Soundararajan, Ole Agesen. VMware Inc.
----
(*) We note that VMware's original goal - to virtualize the x86 architecture - was widely believed to be impossible, but then it was done! As an industry, and with academic research supporting us, let us never allow "impossibility" to get in the way of what must be done.
ref1: https://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/company/vmware-infrastructure-virtualization-leads-the-way-in-reducing-the-carbon-cost-of-growth.pdf
Fred Douglis
Dear Dr. Chien:
This article makes a number of excellent points, and I agree completely that the environmental impact of computing needs to be addressed in numerous respects. One area that was not touched on in this piece is the excessive energy demands of some "proof-of-work" blockchain protocols, most notably Bitcoin. In fact, there are websites and academic literature devoted to this issue, such as https://digiconomist.net/bitcoin-energy-consumption. That site reports that the energy consumed for a single Bitcoin transaction could power nearly 14 U.S. households for a day.
There are numerous blockchain protocols that do not require extensive and expensive computing effort, including some that are trying to serve as efficient brokers for financial transactions. If we want a reduced environmental impact with respect to computing, enabling such technologies and moving "real world" users toward more energy-efficient alternatives will be essential. DARPA recently held a workshop on consensus protocols, with topics including incentivization of participation in distributed consensus without monetary rewards.
Blockchain has many great uses. But, if we want to answer this call to action, lets address the elephant in the room -- cryptocurrency speculation via mining consumes enormous amounts of energy with very significant and damaging environmental effects.
Fred Douglis
Displaying all 2 comments