• Editor

Why data centres aren’t as bad for the environment as you may think


Data centres are often criticised for their carbon intensity and the amount of energy they consume. With the exponential growth of the internet in recent years, data centres are now a fast-growing industry.


They have come under increased focus during the Covid-19 crisis, as people realised how essential they have become to everyday life. In fact, governments now refer to these real estate assets as “mission critical”, as they are vital to the effective functioning of society.


Less than five years ago, articles in the media were warning that digital infrastructure, such as data centres, could consume 20% of global energy by 2025. This was easy to believe against the backdrop of rapid growth in connectivity that was taking place.


Projections from the International Energy Agency (IEA) suggest that internet traffic will double between 2019 and 2022, with the Covid-19 pandemic adding to this trend through greater demand from areas such as online gaming, video conferencing and streaming.


Image: Green data centres: Beyond Net Zero. Rising importance of data centres – the ‘invisible’ assets at the heart of the digital economy


So, with this exponential growth in all things digital and data related, it is still easy to believe that digital infrastructure might be one of the key consumers of global energy. However, the reality is somewhat different from the dire predictions.


Many would be surprised to hear that the data centre industry only consumed 1% of global energy in 2020, according to figures from the IEA. This is a significantly lower figure than the relatively recent predictions of 20% of global energy. And while data consumption has grown rapidly over the past decade, the amount of power that data centres consume has remained largely static.


Image: Environmental Impact. Historical energy use.



The main reason why these predictions were so wrong was the innovation that took place in three key areas. Firstly, the operation and construction of data centres; secondly, the transferral of private data to the cloud; and finally, the design of IT equipment. By analysing each of these areas in more detail it is easier to understand why energy consumption by data centres has managed to remain static despite the data boom.


Data centres are operated more efficiently


Some specialist data centre operators have developed several tools to maximise the efficiency of data centres that they manage. These companies include Switch, Equinix, QTS, Interxion and Next DC. These companies are increasingly setting the global standards for construction, cooling, power optimisation and security.


Structural shift to the cloud


Previously, companies ran their data servers from their own offices, often locating them in the basements. This was a very inefficient use of their infrastructure. As an example, servers would be running 24 hours a day irrelevant of whether they were being used or not. However, companies have now realised that they can cut costs and improve operational efficiency by moving their data to professionally run data centres.


The best run data centres will use artificial intelligence to establish precisely where and when power and cooling is needed Therefore, it is an easy decision for companies to make and it has been estimated that businesses can reduce their energy usage by 30% by using a professionally managed cloud service.


Innovative IT equipment


New advances in artificial intelligence and IT equipment, such as servers, has resulted in data consumption having a far smaller carbon footprint than in the past.


Despite the progress, more needs to be done


For the industry to continue to grow in a sustainable way, data centre companies must actively work towards having a net zero impact. Within the industry, there are leaders and laggards. Although there is a lack of standardised reporting metrics to effectively compare companies, there is one metric that is commonly cited.


Image: Green data centres - Seven key investment areas

</