In recent years, the increase in the number of data centers has been a cause for concern. Energy-hungry, data centers have become essential to the smooth running of the Internet by allowing the storage of an astronomical amount of data in the cloud. If the alarm bell has already been rung, the balance sheet is not as catastrophic as one might think. Nevertheless, the deployment of 5G will present various challenges in the coming years. Fortunately, many innovative solutions are on the table to optimize or even reduce the ecological footprint of data centers.
Environmental impact of data centers to be nuanced
There is no shortage of articles when it comes to pointing the finger at data center energy consumption. Greenpeace had already alerted the general public in 2011 with a comprehensive report on this topic: "How dirty is your data". The figures and statistics are numerous and vary depending on the source. More recently, however, the subject was the subject of a vast study published in early 2020 in the American scientific journal Science. This study covers the years 2010 to 2018.
One of the interesting observations of this article is that the energy consumption of data centers has not increased proportionally to their development. Although the computing capacity of the data centers increased sixfold during the period under study, the energy consumption only increased by 6%. Over the same period, Internet traffic would have increased tenfold.
The authors of this study, members of prestigious American universities, believe that this lack of connection is due to improved technology. Concepts such as Virtual Machines or the localization of data centers considering environmental factors would have helped limit the growth in energy demand. Some of the scientists who wrote the article admitted that they were surprised by the conclusions of the study, which went against the conventional wisdom.
The International Energy Agency (IEA) seems to corroborate this study in certain aspects. According to their analysis, data centers would represent 1% of the world's energy demand. Data transmission networks would also account for 1% of this demand. These figures are lower than those previously advanced.
Environmental challenges to be met
Admittedly, the balance sheet as currently drawn up is less black than expected. However, this does not mean that we should sing a chorus of praise. In the years to come, the use of technology will follow an exponential growth curve. The problem of the environmental footprint is not just a problem for data centers. The IEA estimates that Internet traffic will have doubled by 2022 to reach 4.2 zettabits per year. At that level, there is no point in trying to represent the number of books or soccer fields that this astronomical amount of data can fill.
In the process, the Internet of Things will explode. It is estimated that the number of connected objects will double by 2025 to more than 25 billion units. Moreover, while 4G is clearly less energy intensive than 2G or 3G, the environmental impact of 5G is still controversial. However, its deployment is well underway given the benefits of the technology.
The issue of reducing energy consumption in data centers is therefore more topical than ever. Fortunately, there is no shortage of ideas and innovations in this area.
Reducing energy use in data centers, the technology aspect
The most obvious way to reduce energy consumption in data centers is to improve their operation and the technologies used. Contrary to popular belief, titanic data centers, also known as hyperscale data centers, would be the best students in this area. Indeed, large enterprises and web giants have the resources and technologies necessary to improve energy performance.
If we take the example of Google, the New York Times reports that the California company's data centers have multiplied their computing power sevenfold in five years without using more electricity.
The development of Virtual Machines also helped reduce energy consumption, as well as the growth of services such as PaaS, SaaS, and IaaS.
Reducing energy use in data centers, practical examples
In addition to technological developments, there are many solutions, some of which may be incongruous, to reduce the energy consumption of data centers.
The first consists of natural cooling or free cooling. In particular, this means using free cooling sources. One example is locating data centers in cold locations to limit air conditioning needs. Microsoft has also made a lot of headlines with its underwater data center. These solutions, although commendable, are still harmful to the environment.
Other proposals aim to create a virtuous circle. Stockholm, for example, hopes to provide around 10% of its energy needs through data centers by 2035. Some examples also exist in France, notably in Seine-et-Marne.
Of course, updating equipment also allows us to benefit from new technologies. However, we have to consider electronic waste or other collateral damage.
Finally, artificial intelligence promises great advances by creating smart data centers that adapt to external conditions in real time.
An essential change in mentalities
There is no shortage of solutions when it comes to reducing data center consumption. Nevertheless, the problem remains. The storage of useless data is a menace. In the age of big data, gigantic data lakes are created to store data "just in case". While this approach has many advantages, you also need to know how to empty the data lakes at regular intervals.
While storing unnecessary or unusable data is bad for the environment, it also comes at a cost. The phenomenon has a name: databerg (contraction of data and iceberg). According to the latest report on the subject, only 19% of the data stored by companies would have a real use. 28% of them would be obsolete, redundant, or useless and 53% would be dark data, i.e. data whose usefulness has not yet been defined. In a report published in 2016, the company Veritas estimated that the storage of this dark data and superfluous data would cost about 3,300 billion dollars in 2020. While these figures should be taken with caution, they do indicate the magnitude of the problem.
It is up to each company to define an optimal and effective data management strategy. It is up to each company to define an optimal and effective data management strategy, considering its needs, priorities, and research. Ryax offers a data processing platform that allows to unify the data policy.
The Ryax Team.