The high demand exerted on cloud applications in 2020 directly impacted the data center industry, and it happened immediately. As remote work, virtual meetings, and social distancing became the way of life for many, 96% of data center professionals in a recent survey said the demand on their data centers skyrocketed in 2020.
With people increasingly relying on cloud applications to work and live, maintaining data center uptime while accommodating upticks in network and capacity demand has become more mission critical than ever before. The significance of reliable operation becomes even more clear when you consider that just a millisecond of interrupted power can result in unplanned downtime that can cost up to $17,244 per minute.
While scaling to meet increased demand is essential, it can present challenges. Often at the center of the scaling discussion is the power architecture in place. While AC power architectures are ubiquitous, more than half of data center professionals would consider a distributed DC power architecture as a right-fit solution for data center power system upgrades to meet current and future demand. So, with so many operators actively considering DC power architectures, what’s the holdup when it comes to widespread adoption? This article will explore some of the challenges, how they can be overcome, and dispel a myth or two along the way.
Built to Scale: DC Power’s Potential for Data Centers
The biggest benefits of distributed DC power architecture include saving energy, space, and costs
In a data center, each time power is converted from AC to DC, DC to AC, or stepped down from one voltage to another, significant energy is lost (mostly in the form of heat).
Switching to a DC power architecture from a centralized AC version helps eliminate some of the power conversion steps, resulting in less energy lost and less heat generated. In turn, this supplies higher-quality power and more efficient energy. This supply of efficient energy, coupled with the ease of integrating DC power architectures with on-site energy sources, like solar cells or fuel cells, that produce DC power, can also help facilities use distributed DC power architectures to reduce their environmental footprint.
Distributed DC power architectures can take up less space than AC power and be deployed in different locations throughout data centers. If an AC UPS is at or near capacity, there are limits on how much facilities can scale without having to add another UPS and subsequent equipment. So, without adding another UPS, there may be valuable white space on the data center floor and in the power room that essentially becomes “stranded.” If the power infrastructure were able to scale more easily, that space could be utilized for additional server racks and necessary cooling equipment — in turn, increasing computing capacity.
In distributed DC power architectures, on the other hand, power is scalable at the cabinet level, which allows for the power supply and backup energy source to be stored directly within each server cabinet with the computing equipment. This makes scaling a breeze — buying power and computing capacity with each cabinet as needed.
This level of power scalability, coupled with the ability to recapture otherwise stranded white space, can yield significant cost savings in terms of both capex and opex. So, faced with all of these advantages, why is DC power not the standard power architecture for greenfield data center builds and brownfield upgrades?
Against the Current
One of the biggest perceived hurdles to DC power adoption in data centers is that there are drastically more readily available AC-fed computing and networking equipment options than DC variants. However, while there are certainly plenty of AC servers and routers available, there is also an increasing number of DC options out there today.
Also, in some cases, data center owners, operators, and contractors may not have as much experience with DC power architectures and DC-fed IT loads to confidently pursue a bold DC power structure (or restructuring) strategy.
On top of that, data center professionals have to grapple with a basic human trait: resistance to change. Faced with the frenetic pace of today’s data center buildouts, data center professionals may hesitate to pivot from traditional power architectures, even if its promises outweigh those of traditional methods — especially in mission critical industries that can be notoriously averse to change.
Amping Up Innovation to Grow DC Power
While AC power has been electrical power’s norm since the 1800s, industry leaders are already taking the first steps to make DC power more feasible and accessible. In fact, the International Electrotechnical Commission (IEC) is already devising a plug and socket standard for DC devices, and a standard for plug and socket outlet systems with outputs of up to 5.2 kW has already been approved. But, more work and innovation is needed.
Data centers can take actions today toward adopting DC power architectures. Start small and focus on the server level, where there can be significant savings. Incrementally shifting to DC equipment can help encourage a smooth transition with less inherent risk than a major power architecture overhaul as well as increase confidence and experience working with DC systems.
In the meantime, there will be no shortage of innovation — from automation and increased remote work to 5G and the advancement of IoT/IIoT applications — which will exponentially increase the amount of daily data compute. Optimizing data center scalability is paramount to enabling these emerging technologies and meeting their current and future data demands.
DC power architectures offer a path forward to meeting those demands; we just need to clear the way.