Data centers have long been know to be huge users of energy; the servers, storage and network equipment account for 1% of global electricity demand, which represents a hefty 198 TWh.
With an increasing number of people connecting to the Internet and increasing demand for digital services, and therefore for data center storage, the concern is that the energy use of data centers will spiral out of control.
But a new study released today described such a narrative as "conventional wisdom". Published in Science, it stated: "Several oft-cited yet simplistic analyses claim that the energy used by the world's data centers has doubled over the past decade and that their energy use will triple or even quadruple within the next decade."
These predictions are "extrapolations", said the researchers, that do not account for recent efforts to optimize energy efficiency in data centers. In fact, the study showed that while the amount of computing done in data centers increased by 550% between 2010 and 2018, the amount of energy consumed by the centers in the same period only grew by 6%.
The significant slow-down in energy demand from data centers had already been flagged by a research carried out in the US in 2016. Between 2000 and 2005, estimated the study, electricity consumption in data centers increased by 90%; that number was slashed to 4% between 2010 and 2014.
And yet as every sector of the economy undergoes a digital transformation, data centers have never been more active. "It is important to note that this near constant electricity demand across the decade is occurring while simultaneously meeting a drastic increase in demand for data center services," said the analysts.
In other words, data centers are getting more power-efficient: smart management means that electricity consumption can be better scaled, storage devices are becoming more efficient, and so are networks and infrastructure.
Google, for one, jumped on the opportunity presented by the new research to highlight the company's efforts to reduce the electricity demand in its own data centers. Senior vice president of technical infrastructure Urs Hölzle welcomed the study, saying that it "validated" the search giant's efforts.
He pointed to the company's high-performance servers and the AI-powered cooling system in place in Google's data centers, and noted that the tech giant has deployed smart temperature, lighting and cooling controls to further reduce energy usage. "Compared with five years ago, we now deliver around seven times as much computing power with the same amount of electrical power," noted Hölzle.
According to the research published by Science, one particular trend stands out that is likely to further flatten the energy supply required by data centers. As organizations switch to cloud-based software, most data is likely to be moved to huge, hyperscale data centers – which are more efficient than smaller, local ones.
The growth in hyperscale data centers, designed for maximum efficiency, is perhaps a reason to be optimistic about the future. A report from the Lawrence Berkeley National Laboratory, in fact, estimated that moving all office software applications (such as email or CRM) in the US to the cloud, could reduce the energy used by IT by up to 87%.
That is not to say that local storage is doomed – a prospect that will worry, with reason, enterprises that are attached to keeping their servers on-premise. The industry, indeed, is also working to bring greener solutions to smaller infrastructures.
The Wellcome Sanger Institute in the UK, for instance, processes data about the trillions of cells in the human body via DNA sequencing machines to advance genomic discovery. More data flows in every day, meaning that greater storage, faster connectivity and higher levels of local compute are non-negotiable requirements for the organization.
Simon Binley, data center manager at Sanger, said: "Proximity to the sequencing equipment is a primary consideration for the data center. The bandwidth and latency requirements for the high volume and velocity of genomic data makes cloud services unsuitable."
The Institute, therefore, teamed up with Schneider Electric to build one of the largest genomic data centers in Europe – whilst keeping energy-efficiency and the reduction of power as key priorities. Through smarter management, Sanger's IT team expects to save between 5% and 10% of energy consumption over the first two years that the center is in use, and will then look at increasing the cuts even further.
Overall, the Institute aims to improve the center's power usage effectiveness (PUE) rating from 1.8 to 1.4. Not just to save the planet, but also because of the financial incentive. "Any improvement in PUE and energy efficiency automatically reduces electrical spend and allows us to repurpose investments into other areas," said Binley. The next decade, it would seem, is the one when green goes from fashionable to lucrative.