CyrusOne’s Napier addresses efficiency complexities at DICE South
“In the data center industry, efficiency is treated as some monolithic ‘thing’ that can be easily defined, baselined and drawn down from,” CyrusOne Vice President of Operations Chris Napier told the audience at the Bisnow Datacenter Investment Conference & Expo Southwest 2022 (DICE South) in Dallas-Fort Worth.
On Oct. 22, he joined Schneider Electric Vice President of Solution Architects Joe Reele, ChilliRack CEO and Founder David Klein, Opus Interactive CEO Shannon Hulbert, and Johnson Controls Global Director of Applied HVAC Mukul Anand on DICE’s “Maximizing Efficiency While Minimizing Costs: Advancements in Infrastructure Engineering and Operations – What Remains Inefficient?” panel.
Efficiency, he said, can be difficult to boil down.
“We recognize reduction of consumption as efficiency, but there’s efficiencies of capital deployment, efficiencies of the human capital, efficiency on procurement and efficiency of contract writing,” Napier continued. “To focus on what we do, first comes uptime. What do we do outside of maintaining uptime to increase efficiency and reduce power consumption? Probably universally, I think all of us would answer it’s the ratio between cooling and the critical IT power that we deliver. All of us think of cooling as probably our biggest bang for the buck when we look at where we can extract some efficiencies.”
But focusing on cooling isn’t as simple as installing a widget and then measuring impact, he said. There must also be discussion about labor shortages and other obstacles.
“We can’t necessarily buy the most efficient piece of equipment because it’s probably not available until about two years after we’ve deployed because of supply chain issues,” Napier said. “So, we’re focusing mostly on that interaction between the systems we have, understanding the philosophy of containing air and containing the cooling, and then developing better ways to do that. But you must do that in concert. You can’t just slap on a new technology piece and leave your human capital to figure it out.”
To truly achieve efficiency, data centers will have to better integrate technology with human capital.
“One of the biggest things we can do to develop a more efficient running data center is really to close that gap,” Napier explained. “It might be counterintuitive, but it’s finding that balance between new technologies, the way we layer on technologies and the way that we measure baselines. And equally as important, how do we train our folks to interpret that, react to that and still maintain the 100% uptime?”
Reele pointed out efficiency is about the whole ecosystem, not just cooling, power and IT. Hulbert agreed, adding that telemetry data can be most helpful to find inefficiencies to address in those areas and more. But Napier again reminded the panel about the importance of human capital.
“You must have a workforce that understands how to use that telemetry,” he said. “And often it’s different groups of people who develop the technology, run the telemetry programs and have to take that as a prescription on how to drive your efficiency. We have this idea that efficiency is like a goal line we’re going to get to. But that’s fallacy. You don’t reach efficiency – you continually improve. You get to a point where you’re better than your baseline last week, then next week you’ve got to do it all over again. You’ve got to continually improve.”
But because creating efficiency is about the entire ecosystem within the data center, operators must be exceptionally careful of unintended consequences. You can’t make a change without knowing what the other consequences are going to be, Napier warned.
“As we talk about what efficiency looks like in the future as we layer on technology, we have to be very, very careful that there’s a good marriage between the beautiful minds that are creating the technology platforms and the folks who have that anticipated response when something happens,” Napier said. “You must be very careful that you’ve prohibited any kind of unintended consequences. Because at the end of the day, uptime is our critical deliverable and it’s expensive if you mess it up.”
In closing, while addressing a question regarding diverse levels of power and cooling resiliency, Napier suggested hyper-scalers will play a big role in driving innovations toward efficiency.
“Providers like CyrusOne are starting that evolution now,” he said. “You have critical customer load that are on a 2N infrastructure and then some of it’s on a single UPS back infrastructure and sometimes it’s just straight utility. So, there are some developments there and they’ll lead the way. They’ll force the industry there.”