How can data centers deliver the reliability we so desperately need?
At the time of writing, social distancing and self-isolation mandates are in place across the globe, and much of life has, quite literally, moved online. Thanks to the internet, we use video conferencing to check on family and friends; we learn online; we order food, groceries, and other necessities; and we work remotely. Instant messaging apps, e-learning platforms, e-commerce portals and business collaboration tools rely on the internet backbone and the cloud infrastructure to scale and adapt to market needs. Cisco estimated that by 2021, 94 percent of workloads and compute instances will be processed by cloud data centers.
What makes cloud computing different than on-premise traditional IT infrastructure? Cloud is managed, scalable and hybrid. No matter where your applications are, you simply take care of the services you provide to your customers, 24×7. With COVID-19 affecting our lives, we see an increasing awareness about the role of data centers in supporting critical infrastructure such as national grids, oil and gas refineries, chemical plants, and hospitals, to mention a few.
Data centers cannot fail. Operators are challenged to ensure 24×7 service availability, rapidly responding to increases in capacity demand, managing unstaffed sites, reducing carbon emissions, while addressing growing complexity and cybersecurity threats. For those who have experienced an unplanned outage during the last years, the average cost has risen to nearly US$9,000 per minute, according to Ponemon Institute, without considering the indirect effects like loss of reputation, expensive customer claims and high costs of IT equipment restoration.
How to ensure reliability
What is causing these outages? The Uptime Institute revealed that power failures accounted for 36 percent of the largest global public service outages they tracked between 2016 and 2018. The report also highlighted that 80 percent of those outages were preventable. It also found that the human factor remained relevant, despite high levels of automation and control. The quest for reliable power remains a pre-requisite of a data center’s design and operations. These are some of the aspects that we’ll analyze here.
Let’s start with power generation. The dynamics of energy are changing, creating new opportunities for businesses, organizations and communities everywhere to generate their own power supply and run, if necessary, off grid. When you make and manage your own energy, your costs are predictable, you have a direct impact on your sustainability and you’re not constrained by grid availability. Greener than the traditional power grids, local microgrids were made smart and simple to operate. Power generation includes assets such as gas engines, gas turbines and renewables. On-site energy solutions offer a better way to proactively manage power consumption and help transform your data center from a passive consumer of power to a prosumer.
How can data centers make the grid more reliable? As workloads continue to shift to the cloud, greenfield data centers go hyperscale. There are now more than 500 hyperscale data centers globally. And their power footprint per facility is also increasing, which is directly impacting the grid. A modest peak load reduction in a data center, through participation in utility Demand Response (DR) programs or Virtual Power Plants (VPP), would significantly help the grid maintain its reliability. In the past DR has been largely under-exploited in the data center industry, due to lack of research about the risks and benefits of participating in it. But recently, some data center operators in Europe have chosen to participate in DR, achieving considerable energy savings. On the other hand, VPP are an alternative for those operators that cannot participate directly in a DR program. A software platform matches grid needs with the excess capacity of the many small contributors. This means data center operators can monetize assets that are largely underutilized.
How do we reduce risk through measuring power quality? Electronic devices and automation systems are sensitive to voltage variations and dips in the power supply that are often caused by unexpected connection of energy sources, such as renewable energy and large energy consumers. A fault in power supply can trigger a domino effect on business continuity. Constant measurement and analysis of power quality can reduce that risk and detect unexpected events early on. Operators can receive instant notification if issues such as harmonics, transients, sag/swell or voltage disturbances occur. Operators can also create intelligent logic for custom alarms, demand control, automated power factor correction, or load predictions.
Data analysis and reporting can be performed by an energy and power management software (EPMS), which can display waveform captures with a millisecond timestamp; locate the source of violation; and plot the magnitude and duration of all power quality events. Native integration of EPMS within a multi-discipline infrastructure management platform, can generate significant cost savings and improve user experience at the same time. Operators can benefit from a single-pane-of-glass meant for managing and operating reliable infrastructure. It also helps to embrace transparency of data, multi-discipline governance, energy performance optimization and operations improvement.