Neil Rasmussen is senior vice president and chief technology officer at American Power Conversion Corp. in Kingston, R.I. He founded the company in 1981 after graduating from MIT with an MSEE; his thesis was an analysis of a 200-megawatt power supply for a Tokamak Fusion reactor.
The company got its start in solar electricity before becoming a powerhouse in the UPS market in the mid-'80s. Recently, APC was acquired by Schneider Electric SA and has combined operations with Schneider's MGE UPS subsidiary. The new company is called APC-MGE.
How does APC fit into the big picture with Schneider? They're really gearing up on this whole energy-efficiency avenue, the intelligent design of buildings, and they feel that data centers are a very important part of that story. They felt that it was a whole data center story and not just a UPS story, and APC was going the whole data center route and not the UPS route.
You have a passion about modularized, standardized data centers right down to how the electrical panels are set up. Those standards still aren't here. Who will develop them? And when? It's not just the equipment. It's the process of designing a data center. We have to standardize that process ... because ultimately that process develops the specification for what you want.
There's a huge amount of mistakes made in that process. They're very costly. That's why we end up with all of this waste in data centers. People didn't mean to, but they made a lot of mistakes because their process was poor.
How do you specify density? How do you specify efficiency? How do have a rational discussion about a growth plan today? Most people throw their hands up and say we don't know. It can't be like that. If you're building something you'd better have a plan and understand what it means and what its consequences are. Those kinds of things we're trying to standardize.
Aren't there standards bodies working on this? In the case of efficiency, sure. You have the DOE weighing in and the Green Grid. In other areas, our view is we're building our own system with our own Lego blocks, and we're going to open that up at the point where other people can make Lego blocks that fit onto our system. But we're not going to sit back and wait for the industry to come up with a standard Lego block. Things are moving so rapidly we can't sit around and wait for committees to work on it for three years.
So you can't wait for standards to emerge? Look at hot-aisle/cold-aisle layout in data centers, an obvious, well-known technology put out 10 years ago. Every good design firm knows how to do it. But the first standard from The American Society of Heating, Refrigerating and Air-Conditioning Engineers [that included] hot aisle/cold aisle came out a year ago. Ten years ago, everyone knew how to do it. To get a standard out took another nine years. We can't wait for that.
You're proposing that U.S. data centers move to European voltage standards, and you're integrating that feature into your product line. Why do that? It's a big efficiency pickup and it frees up space. Take a megawatt data center and look at the ROI on that. If I could get rid of the [power distribution units], I could pick up 6% of the power capacity of that data center. That's 6% more IT equipment [load] I can put in there. That's a material amount.
Is this for greenfield data centers only? We have two ways of doing it. One is you can build a whole new data center and that's where you get the biggest impact. People are also looking at it when they're laying out high-density zones. But the efficiency pickup is only for the zones we install. You wouldn't rip out existing PDUs [power distribution units]. It's only for new deployments.
In an interview with Computerworld in 2003, you said about a 17-kilowatt server rack, that "no data center on earth can cool that." Now APC's product literature claims its InRow cooling handle up to 70 kW in a rack. What's changed? Well, I don't know about 70. I don't know where that number came from.
When I said you couldn't cool 17 kW, I didn't mean you couldn't cool a single rack somewhere. I meant you couldn't cool it over a sustained area. You couldn't put in a big row of 17-kW racks. Somebody could drop one rack somewhere and they might get away with it. But you couldn't systematically get away with it and you couldn't be assured that you could deploy it. Most people would put it in there and cross their fingers and hope it wouldn't overheat. To stay in that location, you can put 20 kW in and it's going to work, that's what people want to know.
Now you can do almost 30 kW over a sustained area, if you have almost a targeted air conditioner per rack to get to that density. You can do it at slightly lower densities without that amount of air conditioning in there with some of the InRow solutions.
The technology has shifted, but are we going up to 70, 80, 90 kW per rack? I don't think so. I think we've seen the market stabilize in this 20 to 30 [kW] range.
But isn't the ultimate long-term power trend for chip makers and system designers still moving upward? No. What they're going to do is hold the line on the power envelope. That's what's happening with blades. They're going to say here's my power envelope, here's my form factor, give me the computing horsepower given that. The old way was here's the size of a blade, put as much horsepower as you can in independent of the power. That's all over now. No one is designing blades that way or chips that way.
In an industry where working extra hours and forgoing vacation time is de rigueur, you recently started taking most of the summer off. What's your philosophy about that? I started working 20 hours a day in 1981. Right now when I take some time off, it's to recuperate from 26 years of going at 99 miles per hour. It's R&R, trying to keep healthy, get a lot of exercise, that's my focus when I take time off.
I feel that I have to make balanced decisions in my life. I've got kids ... and I spent a lot of time when they were young working, sitting at a CAD plotter at three in the morning, looking at prints. I've got to make a change and that's what I have been doing.
What do you feel most passionately about? I am passionate about trying to improve the world. You can imagine that after the sale of this company that I don't need a job. The only reason I'm here is that I'm just so excited about everything we've built so far, and we have this opportunity to actually save huge amounts of energy. Some of our original vision we had when we started the company was to improve the world. All of a sudden, we have a huge opportunity here to save a lot of energy, improve the planet -- and we're leading it. It's just a great time to be in the industry.
Do you have more power to effect change now -- or less? I think I'm in a unique opportunity here to change this industry. I have an opportunity to steer the total electrical consumption of data centers down in a significant way and make a personal impact on that. In all these years, I didn't have an opportunity to do that. This is a massive opportunity to do something very positive, to reduce the carbon impact on data centers ... If we were able to deploy everything I've talked about in all of the data centers in the country, on the big electrical meter for the country, you'd see the needle move down. That's exciting.
APC has had a fuel cell-powered UPS for two years now. What has become of that technology? We have them out in the field, but it's not some thing we promote. Fuel cells are still way off. We were using it as a backup supply to replace a generator, but generators run so infrequently that their carbon load is low. They're not on all the time. So as a backup source it's not a big thing.
[Creating hydrogen to power fuel cells] generates carbon. There's a lot of misunderstanding about that. If you take methane and separate it to make hydrogen you release carbon. People think that [fuel cell energy] is carbon neutral. It's not. There's not a big pickup there from a carbon impact.
Energy sources are not the answer now. It's reduction of consumption. That's the big answer if you want to solve the energy problem.
You worked with flywheels in the early '80s before founding APC. Other vendors have commercial flywheel products that provide bridge power rather than rely on UPS batteries during a power outage. They keep things going until the generators come online. Why hasn't APC ever developed its own flywheel technology? I certainly won't rule that out. It's one of those things we keep looking at. I worked on high-speed flywheels, and we built one of the first ones at MIT Lincoln Laboratory. I know what it's going to take to bring that technology in, and it's not ready yet. But low-speed flywheels can make a lot of sense.
But there's a shift in batteries coming that might move us toward smaller, more efficient, less environmentally impactful batteries, and that could push flywheels off the agenda. Batteries and flywheels are going to go against each other for a while. I've got nothing against flywheels. MGE has offered flywheels as an option on some of its big systems, and you'll see those every now and then on projects we do. On our mainstream offerings that we push to the core of the market, it could happen.
Editor's note: You can read Rob Mitchell's blog entry about this interview, along with some portions that didn't make it into this story, here.