FEATURE STORY: The Story Of Storage

Facebook's Prineville, OR data center
Facebook’s Prineville, OR data center

By Bill Trüb
From the March/April 2013 issue

Data centers are crucial to operations across countless sectors, from retail to information technology, government to biotech, logistics to engineering. Many consumers, however, have little awareness or understanding of just how massive and expensive these facilities are. Large-scale data centers are known to use the amount of electricity equivalent to small towns and, despite many greening initiatives, some centers release a significant amount of air pollution in the form of diesel exhaust. Furthermore, the amount of security necessary to run a successful data center is enormous due to the highly sensitive information and pricey equipment housed in such storage units. So high are these stakes that the Telecommunications Industry Association has even published a document detailing the minimum requirements for the infrastructure of data centers and computer rooms.

But the business of IT is one that changes quickly. The International Data Corporation claims the average data center is nine years old, which is troubling when coupled with research company Gartner’s assertion that data centers more than seven years old are obsolete. In May 2011, Uptime Institute reported that 36 percent of large companies will exhaust their IT capacities within the next 18 months. Yet according to a “Green Data Centers” report by Pike Research,the global market for green data centers segment of the industry is expected to more than double in size in the next four years. It can be overwhelming to try and keep abreast of these fast-moving, ever-changing, air-conditioned rooms of priceless information.

Greene And The Greening Of Data Centers

In an effort to reduce the shocking levels of diesel exhaust that many data centers emit into the atmosphere, the industry is being proactive in finding environmentally sound operating solutions. Aptly-named Nicholas Greene, writer for www.greendatacenterconference.com, penned “Ten of the Biggest Data Center Trends” at the tail end of 2012. Let’s recap a few of his most notable prognostications.

“Cloud Computing’s still got a long way to go before it’s the world-changing behemoth that everyone predicts it will be, but this year saw more and more organizations finding their way to cloud computing, and loving every moment of it,” writes Greene. “Hybrid clouds took off, and Infrastructure-as-a-Service, Software-as-a-Service and Platform-as-a-Service vendors really came into their own. Unfortunately, the great strides cloud computing made this year are going to have some unfortunate side-effects in the near future. Moving forward; scalability is going to be a huge concern: our current data center infrastructure, powerful as it is, might not be able to handle the increased demands of the cloud.”

Greene continues, “2012 also witnessed the birth of the software defined data center. As a direct result of this, we’ve been seeing an increased focus on virtualization with the configuration of the data center’s hardware dealt with by upper-level software. Software Defined Networking, though still in its nascent stages, has the very real potential to revolutionize the way data centers are operated, with new options for resource optimization, availability, storage, and mobility.”

According to Greene, the push for eco-sensitive options has been a success. “The environment has been getting a lot of love from data center operators this year,” he says. “We’ve been seeing a massive shift towards green computing throughout 2012, with big names such as Apple and Microsoft hopping on the environmental friendliness train. The looming threat of global warming, coupled with the obvious energy savings one accrues as a result of green initiatives (not to mention the good press an organization can receive) have combined to make green IT a near-integral part of data center design.”

Greene gives us the word of the year: “Server racks are becoming denser and denser as many organizations consolidate their data centers in order to save on energy and real-estate costs. Consolidation is the word of the year, as data centers grow smaller and more powerful and energy management turns from a good idea to an integral discipline for data center operation.”

And finally, openness and transparency is where the industry is headed, led by kingpins Facebook and Google. “In April 2011, Facebook founded the Open Compute Project—an initiative which I’m sure that many initially took as a very bad April Fool’s Joke. It wasn’t—and it’s been gaining steam ever since,” asserts Greene. “The notion that data centers should be defined by their software infrastructure rather than their physical hardware seemed novel at the time, but Facebook has demonstrated that it’s got real value. Even organizations that are typically secretive to the point of paranoia, such as Google, have loosened up a bit, giving us some insight into the inner workings of some of their facilities. Maybe one day in the future, Facebook’s ideals will pay off, and we’ll be rewarded with true transparency in data center operations.”

Google Searches, Hits On South Carolina

Speaking of such Internet giants, Google held a groundbreaking ceremony in January in Berkeley County, SC to announce it will expand its operations at the Mt. Holly Commerce Park. The additional $600 million in investment at the site brings Google’s total investment to more than $1.2 billion. The data center in Berkeley County currently houses thousands of servers to support services such as Google search, Gmail, Google+ and YouTube. As Google’s services grows, the company must ramp up its data centers to meet demand.

Google's data center in Berkeley County, SC. Google is using the rainwater retention pond as another means for cooling its data center.
Google’s data center in Berkeley County, SC. Google is using the rainwater retention pond as another means for cooling its data center.

“Today’s announcement is another big win for South Carolina,” says Governor Nikki Haley. “We celebrate Google’s decision to grow its footprint in Berkeley County with a $600-million investment. When a world-class company like Google decides to expand in the Palmetto State, it shows we are providing the sort of business environment that helps foster success.” Many states aggressively pursue data center business through various tax incentives because data centers are often a boon for local economies.

“South Carolina and the Berkeley County community are great places in which to work and grow,” says Data Center Operations Manager Eric Wages. “When Google first announced plans to come to Berkeley County in 2007, we were attracted to not only the energy infrastructure, developable land and available workforce, but also the extraordinary team from the local community that made us feel welcome. Today’s announcement is just a continuation of our investment in the state. Google is proud to call Berkeley County home.”

Google first announced plans for a South Carolina data center in 2007, making an initial investment of $600 million to get the center up and running. In November 2010, Google announced plans to construct a second building at the site, which is now serving traffic.

Google is also involved in supporting science and mathematics programs in South Carolina’s schools. Since 2008, it has awarded more than $885,000 in grants to local schools and nonprofits. It also has helped implement a free, downtown Wi-Fi network in Goose Creek.

“Google has been a great partner, exceeding expectations when the data center was first proposed,” says Berkeley County Supervisor Dan Davis. “They have invested capital, created good jobs and more importantly partnered with local businesses to help them do business better.”

“When our community came together to develop this business park, we wanted to attract leading companies that would establish deep roots and grow,” says South Carolina Sen. Paul Campbell. “Google’s expansion is an example of how Berkeley County can serve the needs of the world’s most innovative and dynamic companies. I hope Google’s growth here prompts other growing businesses to put down roots.”

Facebook ‘Likes’ Oregon

Facebook stores more than 240 billion photos, with users uploading an additional 350 million new photos every single day. To house those photos, Facebook’s data center team deploys 7 petabytes of storage gear every month. But what do you do with an exabyte of digital photos that are rarely accessed? That was the challenge facing Jay Parikh, Vice President of Infrastructure Engineering at Facebook.

The team decided a dedicated data center at its Prineville, OR campus could house older photos in a separate “cold storage” system and would dramatically slash the cost of storing and serving these files. The facility has no generators or UPS systems, but can house up to an exabyte of data.

Last year, Facebook built a 62,000-square-foot data center on its Prineville campus to house its cold storage, which can house 500 racks that each hold 2 petabytes of data, for a total of 1 exabyte of cold storage. Similar facilities will be built at Facebook’s data center campuses in North Carolina and Sweden, Parikh said.

The cold storage data center has no generators or uninterruptible power supply (UPS), with all redundancy handled at the software level. It also uses computer room air conditioners (CRACs) instead of the penthouse-style free cooling system employed in the adjacent production data centers in Prineville.

Most importantly, each rack uses just 2 kilowatts of power instead of the 8 kilowatts in a standard Facebook storage rack. But Parikh said it will be able to store 8 times the volume of data of standard racks. Not many companies face storage challenges at the kind of scale seen at Facebook. But Parikh believes more companies will be confronting these massive storage issues.

“Our big data challenges that we face today will be your big data challenges tomorrow,” he says. “We need to keep coming up with advanced solutions to our storage problems. The most important innovations are the problems people solve before the scale of the problem emerges. I believe big data is one of those problems. And we won’t keep up unless we work together.”

Facebook completed a second huge data center on its campus in Prineville, Oregon in 2012. The facility is similar to its existing 300,000-square-foot data center, Facebook Data Center Manager Ken Patchett announced at a Prineville City Council meeting.

“We believe the construction of the phased expansion of Building 2, and the operation of Building 1, staffing and supplying of the Prineville Data Center will continue to have a positive impact on the Crook County-Prineville economy,” Patchett told the city officials.

The second building in Prineville created up to 450 construction jobs, with the project lasting approximately one year. At the time, Facebook said it would add 10 full-time jobs in Prineville, where it currently employs 54 full-time employees providing building maintenance, security and server maintenance. The Prineville project is Facebook’s first company-built facility, and is optimized from the two-story structure right down to the servers to reflect the company’s vision for energy efficient data center operations.

CoreSite Realty Picks NJ

CoreSite Realty has purchased a 280,000-square-foot building in Secaucus, NJ for a new data center, and expects to invest $65 million to buy the facility and redevelop the initial phase of 65,000 square feet of data center space.

The facility, which will be dubbed NY2, is the company’s first data center in New Jersey and a sign of continuing activity in the northern NJ market. CoreSite already has a site in New York City and the Secaucus facility will mark an important expansion for the provider.

CoreSite is under contract to acquire the building, with the acquisition expected to close in early February. The 280,000-square-foot facility sits on 10 acres of land, which allows additional data center development as the market demands. At full build out, CoreSite expects it will offer 19 critical megawatts of capacity. Construction will start in Q1 2013, with turn-key capacity expected to be available in Q4 2013.

CoreSite intends to ensure the availability of high-capacity and high-speed lit services as well as a robust dark-fiber tether between NY2 and CoreSite’s NY1 location at 32 Avenue of the Americas in Manhattan, enabling CoreSite to provide seamless interconnection across its New York campus.

The company has been aggressively building out data center campuses across America. Focusing on network centric and cloud oriented applications, these data center campuses are network-dense.

“CoreSite’s entry into Secaucus is an important step in the execution of our strategy to extend our U.S. platform supporting latency-sensitive customer applications in network-dense, cloud-enabled data center campuses,” says Tom Ray, President and Chief Executive Officer, CoreSite. “Our New York campus is designed to meet performance-sensitive customer requirements supported by our location at the nexus of robust, protected, low-latency network rings serving Manhattan as well as global cable routes to Chicago, Frankfurt, London, and Brazil. Additionally, customers are able to connect directly to service nodes for Amazon Web Services Direct Connect.”

The Secaucus facility follows the launch of CoreSite’s previously announced 15 data center, located in Reston, VA. CoreSite’s national platform spans nine US markets and includes more than 275 carriers and service providers and more than 15,000 interconnections.

The availability of direct connections to high speed networks in NY2 will be of particular interest to financial firms looking to reduce latency and improve performance. Three network service providers have pre-committed to serve NY2, consisting of CoreSite partners Sidera Networks, Zayo, and Seaborn Networks, each of which provides high-performance network support to the financial services, cloud and network communities.

“The new CoreSite data center in New Jersey fits perfectly with Sidera’s growth strategy,” says Clint Heiden, President, Sidera Networks. “This expansion gives CoreSite customers immediate access to over 40 financial exchanges and the Sidera Xtreme Ultra-Low Latency Network.”

In addition to the new facility, the company also announced an Open Cloud Exchange, an initiative looking to offer a range of cloud services to customers. The Exchange will offer best-of-breed partnerships and services from a broad range of providers. It capitalizes on demand for hybrid infrastructures, letting Enterprises, Managed Service Providers (MSPs) and Systems Integrators (SIs) in CoreSite facilities connect directly, via a single resource, to the cloud service providers of their choice. This provides customers with flexible options to securely and easily connect to all types of cloud offerings.

“We’re building the industry’s premier home for cloud services,” says Jarrett Appleby, COO, CoreSite. “With networks—the oxygen for cloud services—as the foundation, adding the industry’s leading cloud providers will create best-in-class scalability, management, automation, software, and many-to-many exchange capability. The Open Cloud Exchange offers our customers enormous provider flexibility, guaranteed performance, real-time monitoring, and easy management of cloud infrastructure services.”

The initial four best-of-breed partners in Open Cloud Exchange are CENX, Rightscale, RiverMeadow Software and Brocade.

  • CENX will provide its CENX Automated Ethernet Lifecycle Management software specially designed for CoreSite’s Open Cloud Exchange, enabling easy, single sign-on management of Layer 2 cloud infrastructure services and full MEF CE 2.0 compatibility.
  • RightScale, will provide its platform for deploying and manage business-critical applications across public, private, and hybrid clouds. RightScale offers efficient configuration, monitoring, automation, and governance of cloud computing infrastructure and applications.
  • RiverMeadow Software will deliver its automated cloud onboarding SaaS developed specifically for migrating servers and workloads into and between Carrier Service Provider Clouds.
  • Brocade will provide the hardware infrastructure and switching logic at the heart of the Open Cloud Exchange.

Planned future enhancements include the ability to connect to providers across multiple CoreSite locations within the same metro area; connections between customers and providers in various on-net buildings throughout the country; and the Choice between numerous software and services providers to support performance sensitive customer applications through a marketplace portal. The service is available immediately in seven campuses: Los Angeles, San Francisco Bay Area, Chicago, New York, Northern Virginia, Boston, and Washington, DC.

In addition to this monster of a facility from CoreSite, Northern New Jersey has been no stranger to activity these last few months. Internap announced a 100,000-square-foot project in Secaucus last October, its third in the NY Metro region, to address growing demand. With its supply of data center space in northern New Jersey running low, Digital Realty recently announced construction in Clifton.

Apple Blossoms In NC

Apple currently is building huge new data centers in three states, including the North Carolina iDataCenter. Meanwhile, it is leasing large quantities of data center space in California’s Silicon Valley.

Many of the largest cloud computing providers opted to lease new Internet infrastructure in 2012, according to new data from a veteran market watcher. The report highlights the shifting tides in the “buy or build” decision, in which geography and market economics are contributing to a two-tier infrastructure for many of the largest Internet players, with footprints split between company-built data centers and wholesale space.

Apple, Facebook and Microsoft were among the largest consumers of turn-key “wholesale” data center space in 2012, according to Jim Kerrigan, Director of the Data Center Group at Avison Young. Microsoft leased 12 megawatts of new wholesale space in 2012, with Facebook (10 megawatts) and Apple (8 megawatts) not far behind.

The trend is notable because all three companies have recently been building their own massive data center facilities. Facebook has 1.5 million square feet of data center space that is either built or nearing completion, while Apple has finished its huge iDataCenter in North Carolina and is building new facilities in Oregon and Nevada. Microsoft has built its own server farms in seven sites around the U.S. and Europe over the past 5 years.

After years of building huge data centers in remote areas, in 2012 the geographic focus shifted back to historic Internet hubs in northern Virginia, Silicon Valley and Chicago. Apple and Facebook have moved armadas of servers to rural locations in North Carolina and Oregon that offer cheap power and cheap land. Cloud builders will continue to do this going forward, but a portion of their infrastructure must always be housed near the Internet’s key intersections, where they can connect with dozens of other networks. Both land and power are more expensive in these Internet hubs, resulting in different economics for large-scale new construction. That’s why the largest wholesale data center providers have a large presence in these markets.

General Motors Gets Specific In Georgia

General Motors announced plans to hire approximately 1,000 high-tech workers to staff its new Information Technology Innovation Center near Atlanta. The automaker needs software developers, project managers, database experts, business analysts and other IT professionals for the third of four centers in the United States.

“Locating this center in Atlanta makes good business sense,” says GM Chief Information Officer Randy Mott. “We can draw from a deep pool of high tech expertise through the surrounding colleges, universities and talent residing in the area.”

“This Innovation Center is exactly the kind of employer we want in the state,” says Georgia Gov. Nathan Deal. “The information age will be with us for a long time, and attracting companies such as GM that are on the cutting edge of manufacturing and technology is a huge win for Georgia.”

Mott is leading a rebalancing of information technology at GM under which the majority of IT work will be done by GM employees instead of being outsourced, which has been the GM model for most of the last three decades.

“We look to the Innovation Centers to design and deliver IT that drives down the cost of ongoing operations while continuously increasing the level and speed at which innovative products and services are available to GM customers,” Mott says. “The IT Innovation Centers are critical to our overall GM business strategy and IT transformation.” The location of the fourth site will be announced at a later date.

Gartner’s View On Cloud Computing

Drue Reeves, Gartner’s Vice President and distinguished analyst, recently outlined five trends that will transform the data center industry for Computer Weekly. Reeves’ expert predictions focus heavily on cloud computing, which requires the use of computing resources (both hardware and software) that are delivered over a network, usually the Internet. The name comes from the use of a cloud-shaped symbol as an abstraction for the complex infrastructure it contains in system diagrams. Cloud computing entrusts remote services with a user’s data, software and computation. Here are Reeves’ five trends, in his own words, for the future of data centers.

  • Hybrid IT: Perhaps the greatest effect of public cloud computing on IT concerns operations. IT organizations realize that not only do they need to compete with public cloud service providers (CSPs), but also act as intermediaries between internal customers and all IT services (internal or external). IT organizations are becoming brokers of a set of IT services hosted partly internally and partly externally — that is, of hybrid IT. As intermediaries, IT organizations can offer internal customers the price, capacity and provisioning speed of the external cloud, and the protection and security of the internal cloud.
  • Internal clouds: When businesses grow accustomed to consuming IT as a service, IT organizations will be compelled to build internal clouds. Unfortunately, building an internal cloud is hard work and few blueprints exist. Although vendors are building products that will help customers build internal clouds, there is no turnkey solution. IT organizations will struggle to cobble together the necessary pieces to build internal clouds. Nevertheless, building them will be a key data center trend in 2012 because of the need to compete with external cloud computing.
  • Hybrid clouds: Hybrid clouds are connections between two clouds, usually an internal private cloud and an external public cloud. They are constructed using software that enables applications and data to migrate more easily between clouds. For example, many applications depend on identity management systems to authenticate users, have gigabytes of data, and have input/output latency dependencies for storage. These attributes often prevent applications from migrating to the external cloud, but hybrid cloud solutions them in unique ways. For example, hybrid cloud software can enable WAN acceleration and VPN connections between clouds that allow IT organizations to keep application services and critical data in the internal cloud, and to move the workload itself to the public cloud. As IT budgets continue to shrink and capital resources remain scarce, hybrid clouds will become a more popular option for augmenting IT capacity and enabling disaster recovery than building another data center or signing a long-term outsourcing agreement.
  • User-centric computing: To compete in a global market and retain key employees, organizations often have to accommodate staff who live in remote locations and use personal devices for work. Some organizations are attempting to radically reduce the operational expense of supporting numerous desktop devices for large groups of users with various application requirements. These needs create new challenges for IT organizations to secure data; back up data; support smaller, less functional devices; and support a broader range of devices. Therefore, many IT organizations are rethinking their desktop and mobility strategies and adopting a user-centric, rather than a device-centric, point of view.
  • Data center efficiency: Competing with the external cloud requires IT organizations to strive for hyper-efficiency in their data centers. If critical data and applications are to be housed in an internal private cloud, IT organizations must deliver internal IT services in an efficient, cost-effective manner. This requires them to squeeze further costs out of their data centers by virtualizing as many applications as possible, using storage efficiency technologies such as data deduplication, and buying servers that enable them to maximize space and power and to consolidate applications.