Skip to main content

Paper notes – Energy-efficient cloud computing technologies and policies for an eco-friendly cloud market

Table of Contents

Paper #

Montevecchi, F., Hinterholzer, S., Stickler, T. & Hintemann, R. (2020) Energy-efficient cloud computing technologies and policies for an eco-friendly cloud market: final study report. Available from: https://dx.doi.org/10.2759/3320

Abstract #

The study on “Energy-efficient Cloud Computing Technologies and Policies for an Eco-friendly Cloud Market” is a 16-month service contract funded by the European Commission, Directorate-General for Communication Networks, Content and Technology, which addresses the issue of exponentially growth in energy consumption due to the expansion of cloud services at a European level, covering all aspects related to the underlying technological and structural base. An overview over the relevant technological developments, as well as current voluntary and regulatory policy instruments are developed and provide insights into potentials for a reduction of energy consumption as well as suitable policy options (including Green Public Procurement) to foster eco-friendly efficient cloud services and an energy efficient data centre and network landscapes.

Notes #

  • The executive summary provides a good overview of the key findings of the report, which highlights that a lot of work has been done on data center energy but there remain major gaps in the analysis of computer networks, data storage, and software development in general.
  • The context for the report is a major increase in the energy consumption of data centers in the EU28: 53.9 TWh/a to 76.8 TWh/a between 2010 and 2018. This is expected to grow 21% to 92.6 TWh/a by 2025.
    • This does not match the narrative in a paper last year which suggests only a modest increase in data center energy expected over the coming years, primarily due to the move to cloud computing and improving energy efficiency.
  • Cloud data centers were responsible for 10% of data centre energy consumption in 2010, increasing to 35% in 2018 and expected to rise to 60% in 2025.
  • Computer networks are becoming more efficient over time, but the volume of data is increasing so rapidly that these improvements will be offset by the growth in that data, so total energy consumption will increase overall.
  • Public sector procurement in the EU already considers energy efficiency for many product categories such as buildings and appliances, but there is not a specific process for cloud services. The report notes how energy efficient cloud computing is not currently a consideration in procurement processes.
  • Best practices for data center efficiency already exist and involve widely deployed technologies such as cooling systems, server virtualization, infrastructure efficiency, and use of renewable energy for data centers. No such best practices exist for networks, storage, or software development.
  • “Cisco has identified 24 cloud companies that generate billions of dollars in revenue from cloud services. The data centres operated by these companies are considered hyperscale data centres. Cisco expects the number of these data centres worldwide to increase from 338 to 628 between 2016 and 2021. According to Cisco, 53% of all data centre servers will be operated in hyperscale data centres by 2021 and 55% of data centre IP traffic will be caused by hyperscale data centres (Cisco, 2018a).”
    • Unfortunately, these stats are from a Cisco Whitepaper which is no longer available online. I have seen the same source cited elsewhere but never been able to find the original paper because Cisco don’t maintain a history online. They redirect the original URL to the latest version of the report which doesn’t include this data. This is a major issue with the lifespan of references.
  • The report highlights the wide range of estimates of the annual energy consumption of data centers – 200TWh – 800TWh. The higher values are somewhat extreme (the colours chosen for this graphic are not ideal).

Estimates of the energy consumption of data centres worldwide

  • The estimates for network energy consumption are just as wide – 200TWh – 600TWh:

Network energy estimates

  • These graphs become even more absurd when future projections are plotted. There are a lot of studies which make extreme assumptions based on simplistic extrapolation:

Data center energy projections

  • The report re-uses methods and data from earlier Borderstep Institute papers on IT energy consumption from the same authors. This is important because most data center estimates are based on data published by Koomey et al over the last 10-15 years. It is good to have several independent methodologies so that figures can be compared. For example, in the graphs above, Andrae uses a baseline from Koomey, and The Shift Project also uses a baseline from Andrae.
  • A discussion of edge data centers is factored into the projections of future energy consumption because the report suggests that 40% of data center energy consumption will be from edge data centers by 2030.
  • From a technology perspective, Moore’s Law has been the most important trend in energy efficiency over the last decade, and PUE improvements have been most important for data center energy. Suggestions that both of these are coming to and end are concerning.

Conclusions #

The report is a detailed review of the major issues across sustainable computing and what the EU public sector should do next. It is right to highlight the need for buyers to start asking questions about energy efficiency in the procurement process for cloud services. Until those questions are asked, vendors will feel no pressure to invest in reporting and transparency.