Digital services / Cloud computing

The use of streaming services, video conferencing and cloud gaming has increased considerably, partly due to the pandemic. In March 2020, the world’s largest internet exchange point located in Frankfurt/Main (DE-CIX) recorded a peak throughput value of 9.16 terabits/second (TBit), a volume equivalent to the simultaneous transmission of more than two million high-definition (HD) videos and the highest ever recorded at the site. What is it doing to the climate and what are the carbon footprints of video streaming, video conferencing and online data storage?

The results of Green Cloud Computing (GCC) research report can be used to answer these questions (in German). We developed a methodology to record the environmental impact of abiotic resource depletion potential (minerals and fossil raw materials), global warming potential, cumulative energy demand (energy consumption for production, use and disposal) and the water consumption of cloud services. An Excel tool makes it easy to calculate and evaluate environmental impact. The GCC methodology was applied for the four cloud services online storage, video streaming, virtual desktop infrastructure and videoconferencing.

The exemplary investigation of online storage revealed that the service is offered at varying degrees of efficiency. The global warming potential (GWP) of online (cloud) storage of the four data centres investigated ranges between 166 and 280 kilogrammes of CO₂ equivalents per terabyte of stored data per year. The transparent calculation method and data storage now also allow a detailed analysis of the cloud service to determine the reasons for the difference in efficiency (in this case 60 percent).

The results of the case studies video streaming and videoconferencing provide the first-ever real data collected in practice about the environmental impact of these cloud services.

One hour of video streaming in high-definition (HD) quality (two gigabytes of video material) at the data centre had a carbon footprint of 1.46 grams of CO₂ equivalents. This corresponds to electrical power consumption of around 2.3 watts per video stream, not counting power consumption for data transmission and local IT.

The result of the case study video streaming allows the conclusion that the power consumption and CO₂ emissions in cloud data centres are far lower than previously assumed. The largest share of total greenhouse gas emissions can be traced to home networking and end devices, with size of TV set on which video is streamed playing the biggest role. In simple terms, the bigger the screen, the worse the carbon footprint.

The greenhouse gas emissions of a data centre participating in one hour of videoconferencing add up to 2.27 grams of CO₂ equivalents, which corresponds to electrical power consumption of about five watts per person. In addition, there is the power consumption of data transmission, local infrastructure and the end device. Participating in a one-hour video conference with a laptop is associated with emissions of 55 grams of CO₂ equivalents, roughly the same as travelling one kilometre by train. A large video display increases greenhouse gas emission but videoconferencing is still usually better for the climate than traveling to a physical meeting – unless one travels there on foot or by bike.
The diagram below shows data for videoconferencing via laptop, computer plus monitor or video monitor compared to travel with various modes of transport.

Greenhouse gas emissions of videoconferencing

using different display devices, in passenger kilometres on various modes of transport

Source: Umweltbundesamt

The Green Cloud Computing research project  (in German) analysed the carbon footprint of a computer workspace equipped with virtual desktop infrastructure (VDI) as compared to the classical solution. VDI means that the entire computer desktop in the data centre has been virtualised. All the specific settings, software products and data are no longer installed on the local computer but on a server in the data centre. This no longer requires the standard PC and energy-saving thin clients can be used instead, which serves as the interface to the server. Our calculations conclude that a workstation equipped with a VDI (thin client computer) generates 33 kilograms fewer emissions than a workstation using a notebook or a desktop PC. The calculations include the manufacturing costs for the hardware in the data centre and for the local thin client computers as well as the power consumption in the data centre and at the workstation. However, the sensitivity analysis shows that it is not always more favourable to the climate to move IT services to the cloud. What is important is the equipment of the local workstation, how IT is utilised and whether the infrastructure of the data centre is geared to requirements.

The Green Cloud Computing (GCC) methodology developed by the project enables a first-time calculation of cloud services (raw material consumption, global warming potential and water consumption). The GCC indicators are well suited to communicating and comparing the environmental performance of individual cloud services for customers. They are also suitable for cloud providers to determine the potential to improve the cloud services they provide.

What is remarkable about the GCC methodology is that it provides enough flexibility to interpret the data from different cloud services and to provide answers to various issues in this context. In addition, the method is transparent and easy to use in practice.