Data overload: DC power may be the new normal for efficient data centers

A recent report on DC power revealed some interesting trends related to data center operations and power. Learn what data center conditions look like today and going forward.

Data center with wires

Image: simonkr/E+/Getty images

Data center solutions provider ABB Power Conversion recently released a detailed report regarding data center power and operations. Here are some key findings:

  • 96% of U.S. data center professionals reported that demand on their data center increased during 2020. 
  • Cloud-based applications for financial and banking services, machine-to-machine Internet of Things applications and telecom connectivity were key drivers of the demand increases put on data centers in 2020. 
  • Just over a third of data center professionals surveyed who have seen an increase in demand in 2020 leveraged stranded white space, while a quarter increased usable rack or floor space. 
  • Service and maintenance is the top data center power concern for 2021, followed by power usage and scalability. 
  • Half of U.S. data center professionals are likely to consider power system upgrades to meet increased data center demands in 2021 and beyond. Fifty-two percent of data center professionals consider DC power as a right-fit solution for upgrading power systems. 
  • A quarter of data center professionals are unlikely to build new space and will instead try to maximize their current facilities to meet demands.

I spoke with Jeff Schnitzer, president at ABB Power Conversion, to discuss the report further.

Scott Matteson: Did dedicated cloud data center usage increase in 2020?

SEE: Snowflake data warehouse platform: A cheat sheet (free PDF) (TechRepublic)

Jeff Schnitzer: As seen in the report, 96% of U.S. data center professionals reported that demand on their data center increased during 2020. While we can't say for certain that dedicated cloud data center usage increased in 2020, it's likely that all types of data centers were utilized to support a world gone remote.

In fact, according to an IEA report, between February and mid-April 2020, global internet traffic surged by 40 percent. Further, remote collaboration tool usage surged, with Slack seeing customer growth of 35 percent in 2020 year-over-year. Today, one thing appears clear: Business, and life, is going to look a lot different moving forward, as digital tools such as video conferencing and virtual services like telehealth have become essential parts of our lives. And with this increased reliance on digital tools like these, the cloud itself—and the solutions it enables—has become more mission-critical than ever.

Not surprisingly, the impact of this digital transformation amplifies a broader set of challenges for data center operations. First, how can data centers maintain resiliency, efficiency and reliability amid the unprecedented and ever-growing demand for cloud-based, business-critical applications? Second, how can data centers ensure their facilities are well prepared to scale operations in response to rapidly changing and continuously increasing capacity demands? Responding to these challenges requires that data center operators attend to the critical role power architectures play as potentially enabling or limiting factors for expanding data center capacity.

Scott Matteson: What are some examples of service and maintenance issues or problems and recommended solutions?

Jeff Schnitzer: Data center maintenance can vary depending on the type of data center environment and infrastructure.

Regardless of these variables, though, data center service and maintenance are essential. Not only to ensure reliable, 24x7 operation but also as a long-term cost-savings measure and a way to make operations more efficient and sustainable.

The best way ensure uptime is to work with a trusted partner that knows your equipment inside and out, your facility and its needs, and understands your customers' requirements as well as trends that may influence future challenges and needs. Looking at power equipment through a life-cycle lens instead of a one-time install will help ensure the power solutions can meet a data centers' current and future needs. This includes consulting with the power equipment provider early on in the build process to evaluate the requirements and challenges at hand and leveraging their expertise as well as their access to spare and replacement parts and equipment to keep systems up and running 24x7.

SEE: Kubernetes: A cheat sheet (free PDF) (TechRepublic)

Because at the end of the day, that's what matters most. It may also make sense to look at service-level agreements as part of your initial investment (along with equipment) to help provide added peace of mind—and to best optimize the long-term operating expenses of your power system. 

In addition, there are many technologies to help extend data center life cycles and help ensure uptime. For example, implementing a predictive maintenance-based service strategy. By utilizing innovative sensor technology and web-based platforms, data center operators can leverage remote monitoring, performance predicting, and they can better control and optimize their assets.

Scott Matteson: "Half of U.S. data center professionals are likely to consider power system upgrades to meet increased data center demands in 2021 and beyond." What sort of investments will that entail?

Jeff Schnitzer: There are a lot of factors to consider when looking at the investment that might be needed, especially since data centers have their own unique set of needs. Investments will depend on current infrastructure, current and future demand, and capacity projections, and if operators are looking to upgrade or supplement existing facilities and equipment or if they are considering new, greenfield builds to meet future needs.

Data centers are increasingly reliant on on-site backup power systems to help ensure reliable, consistent power, 24/7, as even a millisecond of downtime can be detrimental to their and their customers' operations. There are a number of options to consider when it comes to power upgrades, and the right choice will ultimately come down to what will best meet their specific needs.

For instance, many data center operators are looking for ways to add redundancy and scalability, especially in the aftermath of the 2020 demand spike driven by the global pandemic, while also remaining highly focused on efficiency and sustainability. So, we're seeing increased interest in distributed DC power architectures, which can directly connect the protected AC feed to the IT rack or cabinet, where smaller, more power-dense batteries and rectifiers are housed directly inside along with the IT equipment. This approach helps reduce the number of power conversion steps required to step down utility power to the servers and routers and enables a more modular, scalable approach to meeting increased capacity demands.

Scott Matteson: "A quarter of data center professionals are unlikely to build new space and will instead try to maximize their current facilities to meet demands." Do you have any recommendations for how they can go about this? What are the unique advantages of distributed DC power architecture?

Jeff Schnitzer: These past 12 months have spurred record spending on data centers, and we don't see that slowing down anytime soon. New research shows the total investment in data centers is expected to increase from $244.74 billion in 2019 to $432.14 billion in 2025.

For data center operators, investing in the right power architecture can make or break the long-term viability, and scalability, of their data centers. Because of this, current and future power architectures are worth examining when data center operators are looking to maximize current facilities to meet increased demand.

Whether they determine AC or DC as right-fit upgrades to supplement existing power systems, it's important to be mindful of the long-term cost savings on the table from making upfront investments to replacing outdated power technology now and in the future.

SEE: From start to finish: How to deploy an LDAP server (TechRepublic Premium)

While many data centers rely on centralized AC power architectures today, the potential benefits of implementing a DC architecture are becoming increasingly hard to ignore—as is evident by the more than half of survey respondents in our recent Data Overload report that said they are likely to consider DC power architectures as right-fit upgrades to supplement existing power systems.

The fact is, DC is more efficient from a power conversion standpoint. Every time power is converted from one voltage to another, or from AC to DC, it incurs losses, primarily in the form of heat. The fewer conversions the power supply undergoes, the lower the losses and the less heat generated that must be removed from the facility to maintain environmental integrity. This heat mitigation also requires additional energy from HVAC systems to remove it, so reducing losses pays double dividends in direct and indirect energy costs. If data center operators make a holistic shift to DC power, it can lead to greater efficiency and sustainability, and in turn to lower installation, capital expenditure and maintenance costs. Maintenance, by the way, was the top power concern for data center professionals surveyed for the Data Overload report.

The modularity of a distributed DC architecture improves the maintainability of the system with hot-swappable, user-replaceable modules for rectification and battery storage. Additionally, in applications where the UPS is at capacity, resulting in unusable white space, DC power architectures can help reclaim that space as usable with just an AC feed from the source to the IT rack or cabinet.

Scott Matteson: What are the limitations or issues with centralized DC power architecture?

Jeff Schnitzer: There is a perceived limitation in the availability of DC-fed IT equipment; however, in the last few years, IT equipment (routers and servers) manufacturers have gotten on board with DC-fed options for the majority of their offerings, due to the many benefits described above.

One of the challenges with the traditional centralized DC power architecture is that power must be distributed from a central location at the user voltage. If the system is 48V DC, for example, the cables are very large because of the high currents involved, and losses will typically be high. This is why a distributed DC power architecture often makes sense. The power is not converted to DC until it is close to the equipment consuming the power and can be transferred at a higher (AC) voltage using smaller cables thus incurring lower losses. Alternatively, a higher voltage DC can be used to reduce the loss, however, this voltage will still need to be stepped down to a voltage compatible with the load equipment. Higher voltage (400V) centralized DC systems are being used, and the efficiencies are improved, but there have been some availability and safety concerns with higher voltage DC systems.

Scott Matteson: What is involved with switching from one to the other (work details, expense, conversions, etc.)?

Jeff Schnitzer: The benefit is a smaller capital and operations expenditures budget. That said, greenfield applications far are easier to deploy with DC than with AC and to extract these considerable financial benefits.  

If a facility is already equipped with an AC power architecture, the cost and effort to convert to a DC architecture is prohibitively high, since the distribution equipment and infrastructure would have to be replaced. It would not make economic sense. However, if the AC UPS is already at capacity, additional load can be added by using distributed DC power cabinets connected directly to the utility feed, since they provide their own interruption protection. This is a great way to utilize what would otherwise be stranded white space. A distributed DC architecture can coexist alongside a centralized AC system when its capacity is limited. Installation is as simple as providing a utility feed to the cabinet location.

Scott Matteson: What should IT pros and DC professionals focus on in 2021?

Jeff Schnitzer: IT and data center professionals are well-aware of the exponential growth of data being produced and processed worldwide. If increasing demand is a known entity, the question becomes how can they better build out their data centers to support that influx and scale to meet the advent of emerging technologies with agility and resilience?

Emerging technologies such as cloud computing, the Internet of Things and smart cities are all expanding and continuing to evolve, so ensuring facilities and equipment are ready and able to meet the massive data demands of these applications will be essential in the coming year. If we consider how IoT devices work, it's all about providing low latency and connectivity to transmit and process large amounts of information quickly. Edge computing is one way to navigate and keep that data close to its output source, but 5G networks must first be more widely deployed—thankfully, that reality is nearly upon us.

With data-heavy, next-generation technologies being developed every day, it's going to be more important than ever for data center professionals to look closely at their existing networking and power infrastructures—and consider alternatives like DC power architectures—to ensure they are able to meet the exponential demand increases that are likely on the horizon.

Also see