We won't discover anything new, and we don't believe any reader will be surprised if we point out the incredible increase in information and communication technology (ICT) consumption worldwide in the last 15 years. We also don't think anyone is oblivious to the revolution about to happen with the advent of new technologies and services that will undoubtedly emerge with the arrival of fifth-generation communications.
But perhaps a significant portion of the world's population is still unaware of the magnitude of the impact on energy consumption that this technological revolution may entail. According to recent studies, the information and communication technology sector is projected to consume 20% of global electricity and emit 5.5% of all carbon emissions by 2025.
In the midst of all this frenzy is the data center industry, a key and indispensable component of this technology and the driving force behind this impending revolution. Data centers currently consume over 2% of global electricity, and their growth is expected to be exponential. These facts inevitably push the data center industry to become one of the most important industries in the world in the coming years.
In the face of this situation, data centers must confront a significant challenge. On one hand, they need to meet the growth demands of service providers, and on the other, they must achieve sufficient energy efficiency to be sustainable in an increasingly climate-threatened world.
It wouldn't be unreasonable to expect that in the near future, new regulations may emerge to govern this critical aspect of the data center industry, namely energy efficiency.
The PUE (Power Usage Effectiveness) is the well-known metric that provides a clear indication of data center efficiency. An analysis of the global PUE reveals data showing a stagnation in the rapid decline of the coefficient that occurred until 2013. According to a recent report by Uptime Institute titled 'Uptime Institute Global Data Center Survey 2020,' the global average of this coefficient has hardly changed since 2013, and there was even a slight increase during 2019.
The cause of this stagnation can perhaps be attributed to the fact that many data centers have reached the maximum energy efficiency that their architecture or construction strategies can offer. Many of these data centers will be unable to make drastic changes to their infrastructure due to the investment involved. This makes it essential to use new technologies to break through this glass ceiling and primarily achieve more efficient cooling given the existing infrastructure and architecture, thus increasing energy efficiency.a.
Another important aspect that has been growing exponentially in recent years is the increase in kW/Rack density in data centers worldwide. The rise of supercomputing, driven by new services such as artificial intelligence, IoT, or cryptocurrencies, as well as more 'mundane' issues like the simple lack of physical space, is leading to an unprecedented increase in this density. This increase is inexorably linked to the need for greater, or in any case, more precise use of existing cooling.
Cooling clean rooms is the key to energy efficiency. Essentially, PUE is reduced to the antagonism between computing consumption and the consumption required for room cooling. Achieving maximum cooling efficiency based on the characteristics and capabilities of the existing infrastructure is, for many data centers today, a real utopia due to the complexity of controlling airflow and the correct relationship between temperature and humidity, as well as the challenge of interpreting the data available with current climate monitoring systems.
The general strategy followed by numerous data centers worldwide is to maintain temperature and humidity within the margins indicated by current standards. However, due to the poor monitoring in many of them, the margins of error are so high (in the available 'snapshot' of the climatic situation) that the strategy always tends to overcool out of fear that the lack of information will lead us to disaster. These strategies are obviously incompatible with the concept of efficiency itself.
The importance of having correct and accurate monitoring of your data center is unquestionable, but even more important is the correct interpretation of this data in order to apply the necessary corrective measures and attempt to achieve the maximum of 'minimize changes - maximize impact.'
The technological revolution we are experiencing may also be the answer we need to find the keys to achieving more efficient and flexible air conditioning. This technology is undoubtedly called Artificial Intelligence. This revolutionary technology offers us the possibility to exponentially increase our capacity for monitoring and analyzing the climate information in our clean room.
The ability of AI to infer new data from existing data, as well as its processing power, allows us not only much more precise monitoring but also the ability to perform simulations with a depth and certainty never seen before.
Unfortunately, it's not enough to have the 'cold data' to take the leap, which often means making an investment, however small it may be, for data center managers.
Having the most accurate information is of little use if you don't have the experience, knowledge of new solutions available in the market, and also, but not less important, knowledge of the new operational standards that are gradually gaining momentum in the field of air conditioning. That's why it's more necessary than ever to find the necessary partner on this challenging journey of climate efficiency in data centers.
