Measuring and managing the carbon footprint of cloud applications and infrastructures
In a context of strong growth in IT as a service (with a 17% increase in 2020 according to Gartner) and the ecological transition, digital services in general and the cloud in particular can no longer avoid examining their environmental impact. How to measure (and continuously reduce) the environmental and energy costs of the cloud?
Is the cloud a solution or a problem for the energy transition?
To begin with, as a reminder, the cloud and its different levels of service (IaaS, PaaS, and SaaS) allows on-demand access to resources and has the notable advantage of offering significant flexibility. With the multiplication of uses and a constantly increasing number of users, the cloud and associated infrastructure and components (such as data centers, networks, and terminals) is today being challenged by its environmental impact. This ranges from the use of natural resources, primary energy, and water to carbon dioxide emissions, among others. Environmental impacts have to be measured across the entire value chain from the construction of the building that host IT as a Service infrastructures to the use of the services themselves as well as the management of their end-of-life.
Because while the cloud can have positive effects (helping with resource sharing, accelerating practices, making telecommuting easier, etc.), the increased demand for bandwidth, data duplication, and the increased demand for equipment to use the services are now being questioned as part of a sustainable digital approach.
This approach to measuring environmental impacts involves using specific functional units in order to help guide choices and management of cloud services: architectures, types of platform, suppliers, etc. This also makes it possible to compare technical solutions and enhance the value of strategic choices both internally and externally.
The necessary measurement over the entire life cycle
In order to offer accurate, relevant data, the life cycle assessment (LCA) drawn from industrial methods offers organizations a standardized and realistic view of the environmental impact of their cloud services. It covers the life cycle starting with the extraction of raw materials up to end-of-life recycling or use, and includes energy and water consumption, waste production, manufacturing, distribution, and the use of services.
It involves measuring, for a functional unit (such as one minute of videoconferencing), the impact of different strata and sub-components required to operate the cloud service This includes: data centers (technical infrastructure and buildings), IT infrastructure (servers, network equipment, storage, etc.), software (and particularly application code), user terminals, and the flow of data transmitted over communication networks. The volume of cloud services used or loading and usage rates in relation to overall capacity help weight these results.
To be as realistic as possible, credible use scenarios need to be developed to assess the environmental impacts of the cloud. Thus, for a publisher of a SaaS web conferencing solution, nine scenarios were drawn up looking at the connection place (EMEA, Asia, USA), terminals used, and communication networks used (4G smartphone, personal computer with ADSL, multi-user conference room with ADSL, etc.).
From this data, organizations can develop concrete reports on the environmental impact of the cloud, across the entire value chain and using indicators related to all types of potential environmental effects: extraction or reuse of natural resources, primary energy used, water, or even the amount of carbon dioxide emitted at the various stages of using cloud services.
Environmental management issues of the Cloud: the real objective of the measure
Measurement for measurement’s sake and retrospective analyses are not very meaningful in this context. The objective is to be able to understand the environmental impact based on the uses of the cloud services, to be able to make good decisions to limit those impacts. With this goal in mind, the measurements and associated calculations must be automated as much as possible to be efficient and limit the resources that need to be devoted to them.
In terms of equipment and materials, this helps guide choices based on their intrinsic characteristics but also by the possibilities for optimizing their lifespan, how much they can be repaired, and even recycling. In the same way, it will be possible to orientate its energy choices (consume less, better and more renewable energy) and architectural choices. Finally, the communication aspects should not be neglected, because it is also by mobilizing and involving users, by making what is invisible and virtual concrete, that it is possible to reduce the environmental impact of the cloud across its entire value chain.
A standard for continually improving environmental performance
Accessing data within the scope of a reference framework to understand the state of current situation, make comparisons, and of course make improvements remains a major challenge for environmental management of the cloud. In France, that is the purpose of the NégaOctet project led by several players and notably supported by Ademe, which consists of building a standard method for assessing the environmental impact of digital services. Its primary objective is to provide users with the means to objectively compare the environmental impacts of digital services just like they currently can compare household appliances, for example. NégaOctet, which is still in its pilot phase, aims to rapidly become a standard.
