Analyzing Advanced Power and Cooling at CAPRE’s Great Cloud Summit: Go Micro, Go Modular
SEATTLE, WA – On November 16, Moderator Bruce Myatt, Data Center Business Leader, Arup kicked off the panel “Advanced Power and Cooling: What types of Innovations Are We Seeing Implemented in Data Centers?” at CAPRE’s Great Northwest Cloud Summit by asking three esteemed to talk about the digital transformation. “How are we going to create the kind of power to keep up with this new world we’re entering?” he asked, first looking to panelist Bruce Brady, CEO and Founder of Server Domes to share some insight on his team’s unique solution, and how it might impact the data center equation, whether that’s on the hyperscaler, colocation, or Edge side of things.
“What’s interesting is that the inventor of the Server Dome, which has been in operation for the last five years, was a traditional CTO,” shared Brady. “Just as many CTOs are frustrated, he was frustrated with building data centers that became obsolete. He was frustrated with a lack of flexibility and a lack of concurrent scalability. You have to have downtime, you have to manage loads, and you have to work with dissimilar equipment – not every data center user has the opportunity to put all of the DC boxes at the same facility and make it a homogenous load. So he sat down and envisioned what the data center would look like, and as he says, he skipped a generation.”
According to Brady, at that point, this inventor was labeled a data center heretic. “This is a very conservative industry – and rightfully so. The cost of downtime is very expensive and so is the cost of running a data center, and not just on a PUE level. You can get extraordinary PUEs but still have incredible maintenance and OpEx costs,” he explained. “It’s kind of like a water bed – as you gain on one side, it moves on the other. When you look at TCO, at the end of the year, it still costs a lot.”
Brady then made a prediction that likely no one in the room had a hard time believing. “I think that the industry is going to come under some increased scrutiny,” he forecast. “As the data curve continues to increase, our carbon footprint is going to become more and more significant. So he designed a data center that took that into consideration. Whether it’s the domes which are made of 65% recycled aluminum or the way we get air, which is ambient air with some supplemental cooling. And we have no batteries – we use fly wheels. It’s a negative on the waste stream.”
In other words, Server Domes brings an elegantly simple design to the market place that is flexible, scalable, and simple. “It’s unmanned, it’s lights out, there’s low water consumption, so that no matter where the need is for the compute…the server dome allows for the flexibility of a moving target whether that’s Edge computing or what have you,” he continued. “So that it be reconfigured as emerging technologies come out. It’s carrier-agnostic, it’s equipment-agnostic. It’s a method in order to get highly efficient compute done based on customer parameters.”
Myatt then looked to Brady’s co-panelist, Rick Freeman, Northwest Operations Unit Manager, at M.C. Dean, who has a pretty disruptive modular data center platform, to hone in on one particular segment – the big boys, aka the hyperscalers and major colos. “The hyperscalers are moving at a very fast pace, and the Edge people are right behind them. In some cases, the Edge is even pushing the hyperscalers. But the problem is that you have a lot of hyperscalers that still have legacy equipment that might not meet the requirements [of the Edge providers]. Modular gives you the ability to build something that’s outside of the data center that’s currently there, or have something that can be put in place in a reasonable, short amount of time, from the time that the land is purchased versus the time that the deal has been put in place. You can actually build our system for a data center in a very short amount of time.” Myatt then asked Freeman to expound a bit more on the power aspect of this equation. Are microgrids, for example, useful?
“We work with a multitude of vendors to be able to provide a number of solutions to different clients,” Freeman replied. “We take it upon ourselves to try to figure out what solutions are available. We don’t build our own UPS and our own transformers. We integrate a lot of different systems to provide the best solutions to customers. And that’s really what you have to do….it’s so important to utilize our engineers, our customers’ engineers, to chart a path forward and also address the legacy equipment. If you can address those efficiency issues affecting that equipment, and in a safe way that doesn’t impact your day-to-day operations, that’s a great way to decrease that footprint and bring that efficiency up.”
Finally, Myatt looked to Steve Unger, UPS 3 Phase Product Manager at Toshiba to talk about the value he and his team bring to data centers in terms of energy storage.
“The utility grid itself has holes in it,” replied Unger. “The shift is going to move towards small microgrid systems, which minimizes downtime but also causes problems on the back-up systems for hyperscalers. They’re going to have to have an alternate energy source. That’s where a lithium, SCIB-type of battery will come into play. Because now you can store energy while you’re shifting from one grid to another. That downtime can be detrimental to a data center, even with generator back-up power. It’s going to be the future of the product line, whether it be on the energy storage side to the back-up system.”
Unger then poured praise on a disruptive new material that has been making waves in the digital infrastructure world – silicon carbide. “It allows you to be 98% efficient. That’s incredible if you think about the efficiency of any kind of technology. You have cooling you have to maintain, which is a fan or might be water or another inert type of material. Plus you have control circuitry. And we know that size matters. As we’re talking about data centers, you have a limited amount of space. So how do you maximize that space? Once the buttons push for that size of data center, you have to look at what costs you money and what brings you money, i.e., return on investment.”
“You have to think about maximizing your white space – where you make money — and minimizing your grey space – the energy side,” urged Unger. “Some people are using modular systems, where you take the gray space and put it outside of your data center. Or it might mean that your product has to be small. Making it smaller is a trend that we’re leaning towards in the UPS industry as well.”
Finally, Myatt closed out the panel by asking Freeman to give a prediction about where the industry is going over the next five years. “There’s a lot of money being spent on green energy, and on peak demand — how you can remove having to pay utility companies so you don’t have to pay for peak demand” replied Freeman. “There’s containerized systems for stationary power, to be able to, say, on a hot day in Arizona, when energy is pricey, keep your data center under 400 KW. In that scenario, maybe the lithium product could come into play. You could utilize that battery to supply the DC power, and at night, recharge that battery, and by the morning, have that power readily available to you.”
To be clear, Freeman concluded, there’s a lot of nascent technology out there that could truly change the data center power and energy game. “We’re not even touching on most of it here today,” he clarified, but stressed that being able to go modular is key. “You can even stack modules on top of each other. But we’re building modular systems that are 72 feet long x 72 feet wide and moving them across the country to stack them like building blocks.”