There’s a downside to all of that big data, social media, and Netflix downloading: Energy and carbon. Data centers are massive consumers of energy, which is a huge component of the cost of the services as well as the environmental impact.
The energy component of operating a data center represents between 20%-30%, according to most estimates. Energy powers the servers and cools the facilities. This is an area that UK-based Verne Global has focused on: Building carbon-neutral data center next to a cheap rewnewable energy supply in Iceland, where geothermal energy is prevalent.
Tate Cantrell, CTO, hopped on the phone with the Rayno Report last week to tell us what’s going on in his world of green, highly secure data. Verne, whose data center is built on the site of a former NATO air base in Keflavik, Iceland, counts BMW and RMS, a risk management firm, among its clients. Colt Technology Service group supplies Verne with modular, pre-fabricated data center modules which snap neatly into place in the highly secure, multi-tenant campus. Then clients go to work designing their own data-center architectures, which are often target applications such as high-performance computing.
Cantrell was recently one of three finalists for the AFCOM Data Center Manager of the Year in 2013. He’s got degrees from the Georgia Institute of Technology and Tulane University. Trained as an electrical engineer, he knows a lot about high-performance computing platforms that are powered by green energy. Prior to Verne Global, Cantrell was VP of data center technologies at DuPont Fabros Technology, a data-center REIT. He built the operational strategies involved in the startup of the DuPont Fabros Development Wholesale Data Center business, the first of its kind in the industry.
Read on, as we tapped Cantrell’s knowledge about what’s going on in his world of green data.
Rayno Report: Tell us About how Verne Global has been growing, and a little about your customers.
Cantrell: We got started in late 2011, and our grand opening was February of 2012. That was getting our product launched, with a few customers in a space. We very quickly got BMW in the facility. They did a lot of work. We gained a lot of information about the automotive industry. In 2013, we doubled our capacity. By September we had doubled the capacity of the site. We’re not seeing an end to this stage of growth.
Our story is about the ability to couple low-cost electricity with green renewable energy.
Another interesting customer story is RMS, which has risk management solutions. RMS likes the dynamic of Iceland. Everything is about cores, the cores of computing. Thats how they look at the world — how big of a infrastructure can I build. When they look at the cost of adding Iceland, the cost of [adding] cores is less. They have to be able to scale, and that’s the reason they like Iceland
Rayno Report: What makes it so much it cheaper to scale there?
Cantrell: Energy is at the core of the data center. We think of the data center as a factory. The core is being more and more commoditized, but the differentiator is the variable cost of the energy. If you can package the energy solution that is predictable for 10 or 20 years, you have something that can be more predictive and predict the bottom side of the balance sheet.
Avis analyzed their customers and increased sales by $100M. People are beginning to realize if they manage there computing infrastructure, they can drive their revenue. Over time, every company is going to have to do it. It’s going to stop being a differentiator, and start being a business.
We see it in our customers. BMW comes to our site and says they save 80% on their high-performance computing costs. The i3 [BMW’s new electric car] is built of carbon fiber. Understanding carbon fiber is super computationally intense. They’ve got some terrific data. When you see the presentations by BMW, when they have the crash tests, they will overlay the crash-test simulation and it matches. It’s amazing how right on top of they are. They can really start to tweak and optimize. That’s exciting.
Rayno Report: You’ve talked a lot about energy and scaleability. What are your customers other top concerns — security?
Cantrell: One other area is support and serviceability. We built a customer service team that allows our customers to deploy with confidence. They open a ticket, our technicians are trained to respond. The clients don’t need to visit the site. They send the kit out, they have vendors that put it together, and it stays up and running. These types of logistics are important. The computing resources that these companies are deploying, they don’t want it to become an HR problem.
Rayno Report: You are focused on private clouds, correct?
Cantrell: We don’t provide public cloud serves. We are the infrastructure provider. We manage the contracts with the power companies, we build the buildings, we own the land. We are responsible for the power, cooling, security, and facilitating connections. Our customers are savvy about IT, and if it’s private cloud, they are probably going to supply it themselves. From a data center provide, it’s important to build an ecosystem.
Rayno Report: Tell me a little about security.
Cantrell: The physical security is very important. ISO 27001 compliant. Physical and data security is something we take very seriously. We’re located on an ex NATO airbase. We’ve gone a step further, we’ve put a ring fence around. By the time somebody gets to a cabinet, they’ve gone through nine layers of security. We need to make sure we’re providing security.
Iceland itself has a inherent security because there is an extra step to get into the country itself.
Rayno Report: Tell us about some trends you see, including SDN and NFV.
Cantrell: We watch companies and data center providers becoming more efficient with the physical infrastructure. They are looking at how much carbon they are emitting.
These companies are doing amazing things. Google is pushing their data center metrics — they are only using 15% overhead on their cooling. But I think, really, the next stage is in the optimization in the computational level. What benefit are you gaining in the computational cycles?
I like the movement toward ARM-type chips in the server environment. So you can turn these things all these things and won’t be spinning 50% of the energy when it’s doing nothing. We need to move in an environment in which they are all fully optimized. Virtualization, SDN, all of these things are moving in this direction, but it’s going to take somebody to tie them into a unified solution.
The nice thing about it, when you got somebody that is going to spend that kind of money and unify it and make it efficiency, when wouldn’t they come to a place like Verne Global?
Just to be clear, I’m an outside observer. I like seeing how people tune the factories. I do see there is opportunity for somebody to go in and tweak these things so they are more efficient. You have to break things apart and make them work independently, but they you have to bring them together.
Let’s take the servers themselves. We’re starting to see that the actual hardware, at high-performance computing has specific requirements. You may need things like Fibre Channel connections to storage, because maybe the application is very chatty. You end up tuning the hardware to the application. Let’s build the hardware to the applications.
Rayno Report: It sounds like you see other areas in which data centers and big data are changing business.
Cantrell: The main thing we are really watching is how the high performance nature of computing, when will it become mainstream for typical companies. What is going to be the killer app for a generic company on the Dow, when are we going to see that company leveraging high-performance computing to get better answers to improve their business.
It’s Siloed at this point, you have areas in companies where they are [using big data] to build a better mousetrap. More and more, I’m seeing these areas emerge, I have information about how people interact with their smartphone. All of that is moving in the direction, there are going to be these huge multidimensional analysis of what these companies can do.
Create the big data warehouse, maybe you can run a job four times longer in iceland and you’ll get better answers. How are we changing data and information.
Rayno Report: Thanks Tate.