CyrusOne, the colocation business of Cincinnati Bell, may be spun off or undertake an IPO.

Cincinnati Bell is studying a spinoff or IPO for its thriving CyrusOne colocation business, the company said yesterday.
The move could allow Cincinnati Bell to benefit from investor interest in the data center and cloud computing sector, while shifting significant capital expenses off the telecom company’s balance sheet.

Cincinnati Bell said it would take six to 12 months to make a decision on the best path for CyrusOne, with options including “a partial separation through a sale, initial public offering, or a full separation.” Company officials said the analysis would examine which option created the most shareholder value.

“We expect our data center business to continue its significant growth, and the decision to consider structural, capital and financial alternatives demonstrates our commitment to maximizing this growth and maximizing value for our shareholders,” said Jack Cassidy, president and chief executive officer of Cincinnati Bell. “Our ultimate decision should highlight the value, strong performance, and growth prospects of our businesses while also focusing on opportunities to strengthen the balance sheet of Cincinnati Bell. While this process is underway, we will continue to execute on our existing strategies.”

Cincinnati Bell acquired CyrusOne in 2010 for $525 million, seeing colocation as a potential growth engine. The deal has paid off handsomely. In the fourth quarter, the CyrusOne colo business had revenue of $49 million, up 21 percent from the same period a year earlier. The company’s customers include 16 of the top global 100 companies and four of the top 10.

Building to Keep Pace With Demand

During the quarter, CyrusOne sold 43,000 square feet of data center space, while adding 27,000 square feet of space. The company now has a total of 763,000 square feet of space, which is 88 percent occupied.

That means that to continue growing the business, CyrusOne must continue building data centers. The company is currently working on data center expansion projects in Houston, Austin, Dallas, San Antonio and Phoenix. It has also opened facilities in London and Singapore.

Data center construction requires a significant amount of capital, and Cincinnati Bell says that one goal of its strategic review is “leaving Cincinnati Bell with an appropriate level of debt for its communications business.”

Wall Street Likes Data Centers

Why consider a spinoff or IPO for CyrusOne? Data center stocks have done well on Wall Street in recent months, strongly outperforming telecom shares. Publicly-held colocation and managed hosting providers gaining between 14 and 26 percent in the fourth quarter, while shares of Cincinnati Bell (CBB) were flat.

CyrusOne is a separate legal entity, and has a distinct management team. Over the next six months, Cincinnati Bell’s management and board of directors will consult with its financial and legal advisers to explore “structural, capital and financial alternatives” for the data center business. The company said it doesn’t intend to discuss its progress until the evaluation is completed.

While the company will evaluate alternatives, it said it could eventually decide to continue operating CyrusOne and not pursue any changes.

The exterior of one of the northern Virginia data center properties operated by DuPont Fabros Technology.

Leasing at new data centers built by developer DuPont Fabros Technology has been slower than expected in several key markets, the company said today,  DuPont Fabros said it remained confident in the long-term outlook for the data center market, but that given the current pace of leasing it is lowering its revenue guidance for 2012.

Shares of DuPont Fabros (DFT) were down by about 6 percent in midday trading on the New York Stock Exchange, declining $1.49 a share to $24.01. Shares had been as much as 9 percent lower in after-hours trading last night.

CEO Hossein Fateh said leasing has been slower than expected in New Jersey and northern Virginia, but about on track with expectations in Silicon Valley. DuPont Fabros has recently opened new data centers in each of these markets.

“Leasing in New Jersey has been slower than expected,” said Fateh. “Northern Virginia remains a solid market for us. Although leasing has been somewhat slower than expected in this market, we have no concerns about leasing this space due to the demand.”

Competition in Santa Clara

Analysts have also been focused on DFT’s leasing progress at its site in Santa Clara, California, known as SC1. Santa Clara is a particularly competitive market, and DuPont Fabros has been able to lease 25 percent of the first phase of SC1, including a new 2.28 megawatt lease announced Tuesday.

“We have competition in Silicon Valley, but we also have the majority of the available space,” said Fateh, who said competition had caused a “market disruption” in Santa Clara. In other markets, he said, the slower pace of leasing was tied to internal decisions by customers, rather than compeittion from other wholesale data center providers.

The company’s assessment of the market, shared in its quarterly earnings call, raised a key question for the industry: is the leasing slowdown specific to DuPont Fabros, or a broader change in the demand trend that could affect other companies? We’ll know more in coming weeks, as other players in the data center industry report their earnings.

Will DFT Continue Building Big?

DuPont Fabros builds some of the largest data centers in the industry, which gives the company the ability to rapidly lease space in active markets. But when demand slows, it leaves the company with a larger volume of vacant wholesale space, placing pressure on the company’s leasing efforts. Other players in the wholesale data enter sector tend to build in smaller increments than the 13 megawatt to 18 megawatt phases that DFT brings to market.

Fateh defended DuPont Fabros’ focus on building big, but said the company will adjust its construction plans for the second phases of SC1 and ACC6, deploying new space in chunks of 9 megawatts or even 4.5 megawatts at a time rather than 18 megawatts. But he said the focus on larger facilities is driven by the economics of the data center market.

“We believe that as a landlord operating for the long-term, these are the buildings we want to own,” said Fateh, who said 36 megawatt facilities offer economies of scale not available in 10 megawatt buildings. “We feel very comfortable with our market and our product.”

Emergence of “Super Wholesale”

Fateh said DuPont Fabros was seeing more interest from companies with “super wholesale” requirements of 10 megawatts or  more of critical power. In the past, companies with these large space requirements would be candidates for a stand-alone data center. But now some of these companies are doing comparison shopping between build-to-suit opportunities and wholesale data center space.

“We’re seeing these deals in the market,” he said, noting that these huge requirements were limited to a small number of companies, but represent a substantial new opportunity for wholesale data center operators. Deals of this size could rapidly fill large swaths of wholesale space, with the tradeoff that they might also involve slightly lower returns on capital.

An aerial view of the new DuPont Fabros data center in Santa Clara, Calif.

Wholesale data center developer DuPont Fabros Technology has signed new customer leases at its data centers in New Jersey and Silicon Valley, the company said today.

Two new tenants have signed leases totalling 1.71 megawatts of critical power – about 8,120 square feet of raised-floor technical space – at the NJ1 facility in Piscataway, New Jersey. In California, a single tenant has leased 2.28 megawatts of power (11,000 square feet of space) at the SC1 facility in Santa Clara.

DuPont Fabros also announced that it has opened Phase II of its CH1 data center in Elk Grove Village, Ill. The second phase of the massive facility is 79 percent pre-leased, with 57 percent of those leases commencing on Feb. 1. The opening follows an active fourth quarter in with DuPont Fabros (DFT) opened two new data centers (SC1 and the first phase of ACC6 in Ashburn, Virginia) totalling 31.6 megawatts of critical power.

“Over the past 15 months we have opened five new developments comprising 85.8 megawatts of critical load and leased about half of this space,” said Hossein Fateh, President and Chief Executive Officer of DFT. “Our primary focus in 2012 is to lease our remaining available space. The timing of corporate decision-making by potential tenants to execute leases is not always predictable, but we remain optimistic regarding the long–term demand for our strategically located wholesale facilities.”

Focus on Leasing Space

The pace of leasing at the New Jersey and Santa Clara facilities is being closely watched, as the DuPont Fabros facilities represent the largest chunks of completed wholesale data center space in competitive markets. NJ1 is now 34 percent leased, while 25 percent of the space is filled at SC1.

During the full year 2011, the company signed 14 leases totaling 24.92 megawatts of critical load and 133,716 raised square feet, with an average lease term of 7.9 years and approximate contract value of $428 million.

For the quarter ended December 31, DuPont Fabros reported earnings of $0.12 per share compared to $0.10 per share for the fourth quarter of 2010. Revenues increased 13 percent to $74.4 million for the fourth quarter of 2011 over the fourth quarter of 2010, primarily due to new leases commencing at data centers in Virginia, New Jersey, Chicago and Santa Clara. Funds from Operations (FFO) for the quarter ended December 31, 2011 was $0.37 per share compared to $0.33 per share for the quarter ended December 31, 2010.

DuPont Fabros established an FFO guidance range of $1.31 to $1.51 per share for the full year 2012, establishing a midpoint about 20 cents lower than the company’s full year 2011 FFO of $1.61. DFT said the reduction was due to a tange of factors, including the impact of  lower capitalized interest expenses. No new data center construction is scheduled for to commence in 2012, and the company’s capitalization policy is to stop interest capitalization when projects are placed in service.

After the guidance adjustment, shares of DuPont Fabros were about 9 percent lower in after-hours trading.

The new designation from the Uptime Institute provides third-party assurance that the site management satisfies industry-recognized criteria for 24 x 7 uptime. The evaluation criteria was developed by and for the data center industry and validates staffing levels, a training and professional development assessment, preventative maintenance program and processes, operating conditions and housekeeping, and planning management and coordination practices and resources. To have a site awarded the Management and Operations Stamp of Approval it must be an operational data center, have 24×7 uptime, have a significant cost of downtime, be a non-Tier certified design and facility, and have a commitment to excellence in operations.

Equinix

Equinix (EQIX) announced that  it is the first company in the Americas to receive the new Management and Operations (M&O) ‘Stamp of Approval’ Award. Equinix was a key member of the Uptime Institute’s eleven member M&O Coalition that developed the assessment criteria for its new Management and Operations Program and Site Assessment Service. “Equinix’s exceptional uptime is one of the most important reasons our customers choose us and our work with the Uptime Institute helps us become even stronger in the critical area of management and operations,” said John Shultz, vice president, global product management and development at Equinix. The site award was given to the  Equinix Ashburn data center.

Colt

Colt announced that it has been awarded the Management and Operations Stamp of Approval for its London 3 site. Colt underwent an intensive audit which scrutinized every aspect of how it manages and operates its data center site. The audit took into consideration everything from the processes for servicing equipment and investment in training through to the effectiveness of its communications to staff and subcontractors.
“This is a milestone, not only for Colt but for the industry as a whole,” said Guy Ruddock, VP of Operations at Colt. “This standard is industry-led and directly links staffing, maintenance and operations to uptime and through it we have proved our people and processes are world class.”

“Colt’s participation in the coalition went above and beyond our expectations,” said Julian Kudritzki, Vice President at the Uptime Institute, “not only helping clarify those behaviours that need to exist in a data centre, independent of the infrastructure, but opening up access to staff and facilities for hands-on validation.”

A map of the global audience for Facebook, created by Paul Butler, visualizes the geographic spread of its user base. (Source: Facebook)

Facebook has invested more than $1 billion in the infrastructure that powers its social network, which now serves more than 845 million users a month around the globe. The company spent $606 million on servers, storage, network gear and data centers in 2011, and expects to spend another $500 million this year, Facebook revealed Wednesday in its filing for an initial public stock offering.

The company’s massive armada of servers and storage must work together seamlessly to deliver each Facebook page, the company said. “Loading a user’s home page typically requires accessing hundreds of servers, processing tens of thousands of individual pieces of data, and delivering the information selected in less than one second,” the company said.

Facebook reported in its SEC filing that it owns “network equipment” valued at $1.016 billion at the close of 2011. The number reflects the expense of rapidly building a massive Internet infrastructure, including Facebook’s shift from buying vendor gear and leasing data centers to building its own servers, racks and custom data centers.

Facebook: Investment is Paying Dividends

Although the shift requires huge capital investment up front, Facebook said its custom designs are “ significantly reducing our costs compared to the usage of traditional servers and leased data center facilities.”

But that investment is paying off. In 2011 Facebook made about $1 billion in profit on revenue of more than $3.7 billion.

Thus far, Facebook’s spending compares well with its Internet-scale peers. Google spent $951 million on its data center operations in just the fourth quarter of 2011, with infrastructure capital expenses of  $3.4 billion for all of 2011.

Facebook’s spending is likely to increase as it continues to build out its own data center infrastructure.  The company opened its first  data center in Prineville, Oregon in April 2011, which uses a simplified power distribution system and is filled with servers and racks customized for Facebook’s needs.

More Data Centers in the Pipeline

The social network currently has three more data centers under construction, including a second facility in Prineville and new data centers in Rutherford County, North Carolina and Lulea, Sweden.

The company’s construction costs are also competitive. Facebook spent about $210 million to build 28 megawatts of data center space in Prineville, which works out to about $7.5 million per megawatt. The most efficient providers are building scale-out data center space at between $5 million and $9 million per megawatt. Enterprise data centers, which require additional investment in on-site redundancy and security, can cost $15 million per megawatt.

Facebook said it expects to spend $180 million on real estate leases in 2012, but did not break out how much of that was dedicated to leasing of wholesale data center space, a market in which Facebook is one of the  largest tenants. We have previously estimated Facebook’s spending on data center leases to be at least $50 million a year.

One thing’s for sure: Facebook’s network is only getting bigger. “We plan to continue to significantly expand the size of our infrastructure, ” the company said. “We are investing in additional Facebook-owned data centers in the United States and Europe and we aim to deliver Facebook products rapidly and reliably to all users around the world.”

Cisco Systems says it has updated its Catalyst 6500 family of switches (shown above) for 40 GB Ethernet and 100 GB Ethernet. (Photo: Cisco)

At Cisco Live 2012 London Cisco Systems (CSCO) announced it has updated its switching portfolio with 100 Gigabit Ethernet (GE) and 40 GE capabilities, the next speed limits for networking. These new capabilities provide a holistic architectural approach across campus, data center and service provider environments.

A 40 GE option is available for the Catalyst 6500 switching line and both 40 GE and 100 GE capabilities are available for the Nexus 7000 portfolio.  Two new M2-Series modules for the Nexus 7000 are available: a 2-port, 100GbE Module delivers up to 32 100GE ports and the M2-Series 6-port 40GbE Module provides 96 non-blocking 40 GE ports.

The new Nexus 3064-X switch supports a variety of 1/10/40 Gigabit Ethernet connectivity options as a high-performance, high-density, ultra-low latency Ethernet Switch. A new Nexus 1010-X Virtual Services Appliance offers a dedicated hardware platform for scalable deployment of services critical to virtualization infrastructure.

Providing investment protection for the Catalyst 6500 Series, Cisco is also offering a new 40 Gigabit Ethernet Interface module to help customers prepare their networks for transitions to 40 GE. The Catalyst 6500 E Series chassis are now validated to support full 160 gigabits per slot throughput in active/active mode.

Aimed at space-constrained environments the new Catalyst 4500-X switch is a fixed aggregation 1 RU switch. Cisco also announced simpler network virtualization functionality for its Catalyst 6500, 4500 and Aggregation Services Router (ASR) 1000 product line with a new technology called Easy Virtual Network as well as scalable virtual services with a new Nexus 1010-X appliance for the data center.

Operational Efficiencies

To compliment accelerating Catalyst and Nexus switches to 40 GE and 100 GE options Cisco introduced solutions to help simplify network virtualization and enhance network monitoring, troubleshooting, and management.   New features in Cisco NX-OS software inclue  PowerOn Auto-Provisioning and Python scripting to customize network behavior based on events as they happen, simplify configuration, and harness network intelligence. A Data Center Network Manager (DCNM) simplifies Unified Fabric management and provides a converged LAN and Storage operational dashboard, bringing visibility and advanced diagnostic capabilities into the virtualized data center or cloud.

“Network virtualization is becoming more important as enterprises seek to address security and regulatory concerns without the expense of overlay networks,” said Dominique Massoni, Network Engineering Operations Manager at EDF Group.  There are a number of available solutions, with varying degrees of scalability and complexity. We found Easy Virtual Network to be the right solution for us. It is simple, easy to replicate between sites, and is fully compatible with the solution we have been using, so we didn’t have to redo the work we’ve already done.”

• DuPont Fabros Technology – Wholesale data center developer DuPont Fabros Technology (DFT) today announced that the control procedures for its data centers that required the certification have been found to be suitable in design and operating effectively according to Service Organization Control Reports (“SOC1″).  These were prepared under Statement of Standards for Attestation Engagement No. 16 (“SSAE 16″) by one of the Big 4 independent auditing firms. “We’re committed to operating our data centers and facilities at a level that meets or exceeds the highest industry and regulatory standards,” said Hossein Fateh, President and Chief Executive Officer of DuPont Fabros Technology, Inc.  “We will continue to obtain annual SSAE 16 reports for all future properties we develop, place in service and operate.”
• IO – Modular data center specialist IO said the company has completed a SSAE 16 Type 2 audit covering three operational data centers – IO New Jersey, IO Phoenix and IO Scottsdale. The company said the certification of IO New Jersey and IO Phoenix further validates the resiliency and operational excellence of IO’s next-generation modular data centers. “Designed, engineered and manufactured in the US using standards-based components and processes, IO.Anywhere offers unmatched reliability, in a repeatable manner,” said David Shaw, Chief Operating Officer of IO. “This SSAE 16 accreditation will only further accelerate the increased customer adoption of our highly secure and scalable modular Data Center solution.”

Nimbus announces E-Class Flash Memory.  Nimbus Data unveiled the new E-Class Flash Memory System, which it said ws the industry’s first fully-redundant multiprotocol solid state storage system.  Built for applications such as enterprise-wide server virtualization, Web infrastructure, and database clusters the E-Class platform contains no single point of failures. The E-Class platform consists of a pair of redundant controllers and up to 24 solid state storage enclosures. New features supports 500 terabytes as one logical pool, consumes as little as 5 watts per terabyte, and has comprehensive data management software and “unified” storage without any licensing fees. “The Nimbus E-Class sets a new standard for solid state storage scalability and operating cost economics,” said Benjamin  Woo, program vice president, worldwide storage systems at IDC. “Large enterprises and cloud providers must consider the significant infrastructure consolidation possible with all-flash storage systems. By providing both innovative hardware and comprehensive software, Nimbus is well-positioned to not only capitalize on the need for high-performance systems but also the significantly greater trend towards primary storage based exclusively on solid state technology.” The Nimbus E-Class is available immediately.

SGI introduces modular InfiniteStorage Platform. SGI announced the introduction of SGI Modular InfiniteStorage, an integrated server and storage platform designed to provide cloud and data storage customers with the maximum flexibility for compute and data management applications. To meet diverse data environments the new platform will be available as either SGI Modular InfiniteStorage Server (SGI MIS Server) or SGI Modular InfiniteStorage JBOD (SGI MIS JBOD) products. When built into SGI’s D-Rack cabinet configuration, up to 2.37 petabytes of capacity and up to forty discrete processors can be supplied within a single standard 19 inch rack footprint. ”The SGI Modular InfiniteStorage platform is designed to couple very dense storage and compute capabilities in an adaptable platform, to give cloud and other storage IT customers important new choices for tuning and growing the system to meet their specific requirements,” said Steve Conway, IDC research vice president for HPC. “The SGI Modular InfiniteStorage platform aims to allow IT managers to design customized solutions based on standards-based components.” Both storage products are available for preorder now from SGI.

EMC Isilon scale-out NAS with Hadoop support.  EMC announced that EMC Isilon scale-out NAS with HDFS, combined with EMC Greenplum HD, is an enterprise-ready solution with native support for Hadoop. EMC Isilon’s OneFS 6.5 operating system natively integrates the Hadoop Distributed File System (HDFS) protocol and delivers the industry’s first and only enterprise-proven Hadoop solution on a scale-out NAS architecture. As a single vendor solution it leverages end-to-end data protection including Snapshots, Replication and Backup for Hadoop Big Data. “Hadoop will be a game-changer in unstructured data analytics once enterprise customers are convinced that adoption can simplify their infrastructure and architecture without compromising data security or management capabilities,” said Sam Grocott, vice president of marketing at EMC Isilon. “We’ve taken the guesswork out of Hadoop deployment by bringing together the strengths of Isilon’s enterprise scale-out NAS storage systems with the Hadoop ecosystem. With EMC Isilon, enterprises now have the type of archiving, replication and management capabilities their businesses rely on combined with the innovative analytics of EMC Greenplum and the training, services and support of EMC to make the most of their Hadoop deployments.”

Digital Realty (DLR) now projects funds from operation (FFO) for 2012 in a range between $4.34 and $4.48 per share. That translates to expected FFO growth of 7.2 percent to 11.4 percent over the 2011 range of $4.02 to $4.05 a share. The company said it expected to spend $300 to $400 million acquiring income producing properties, and invest between $700 million and $900 million in redeveloping data center space.

On the leasing front, Digital Realty signed leases for 446,000 square feet of space in the fourth quarter of 2011, representing $50.9 million in annualized GAAP rental revenue. This includes approximately 165,000 square feet of Turn-Key Datacenter space at a rate of $178 per square foot, 87,000 square feet of Powered Base Building space $730 a square foot, 177,000 square feet of Built-to-Suit space leased at an average $84 per square foot, and 17,000 square feet of non-technical space at $18 per square foot. In addition, leases signed for colocation space totaled approximately $2.1 million of annualized GAAP rental revenue, bringing total annualized GAAP rental revenue to approximately $53.0 million in the fourth quarter of 2011.

‘Strongest Performances to Date’

“Lease signings in both the fourth quarter and full year 2011 represented our strongest performances to date,” said Michael Foust, Chief Executive Officer of Digital Realty. “Markets that experienced the highest level of activity during the year included Dallas, Northern Virginia, Silicon Valley, Los Angeles, Amsterdam, Singapore and Melbourne.

“In addition to the quality of our global sales and portfolio management teams, we believe that the strength of our leasing program is the result of our geographically diverse portfolio,” Foust added. “This enables us to deploy growth capital to meet our customers’ data center requirements across multiple markets, while mitigating exposure to any one market.”

Of the total leases signed during the fourth quarter of 2011, approximately 325,000 square feet was for space in the company’s U.S. portfolio. This includes:

• 104,000 square feet of Turn-Key Datacenter space leased at an average annual GAAP rental rate of $151.00 per square foot
• 28,000 square feet of Powered Base Building space at $30 per square foot
• 177,000 square feet of Build-to-Suit space at $84 per square foot
• 16,000 square feet of non-technical space leased at an average $17.00 per square foot.

Digital Realty also leased 114,000 square feet of space in Asia, and 6,000 square feet of Turn-Key Datacenter space in Europe.

Datapipe saw its strongest growth in Europe and Asia, where business grew at an annual rate of more than 50 percent. Adoption of Stratosphere, the company’s Enterprise Cloud Platform, more than doubled for the third consecutive year, the company said. Datapipe also significantly expanded its New Jersey, London and San Jose data center footprints.

“Datapipe has built a team that includes some of the best talent in the industry. Their commitment to delivering value for our clients every day drives our success,” said Robb Allen, Datapipe CEO. “2012 will be another exciting year as we launch new products and enter new markets.”

Datapipe is based in Jersey City, NJ, and provides managed services to enterprise companies, with many customers in the financial services and pharmaceutical industries. The company has more than 1,500 customers in six data centers and eight office locations in the United States, the United Kingdom, and China.