Figure1. 1&1 created a highly reliable colo and hosting environment


 

1&1 Internet rose to the web-hosting industry’s top spot worldwide by offering reliability, innovation, and efficient, automated operation. As a result, the Germany-based company posted a 99.9 percent uptime rating.

1&1 offers dedicated servers that help control costs for business and home website owners. The 1&1 business model has drawn more than nine million customer contracts, which operate on more than 70,000 servers. “We recognized that building data center capability in the U.S. was essential to maintaining our promise of optimal speed, connectivity, and reliability for web site owners here,” said Thorsten Ziegler, head of Data Center U.S. for 1&1.

The company began operations in the United States with a 4,500-sq-ft colocation data center in New York City. The business grew exponentially, and the data center quickly reached its capacity. 1&1 then expanded with another colocation center in New Jersey, which also quickly filled.

Company executives sought a long-term solution that would accommodate continued growth. 1&1 set out to build a new cutting-edge facility, so they retained Burr Computer Environments, Inc., of Houston, to design, build, and manage the project construction. The project began with the leasing of a warehouse in Lenexa, KS, near Kansas City. The area is not only geographically centered, but it’s also situated near a concentration of fiber optic cables running along the many railroad beds that track through the city. The plethora of cables, two redundant backbone routing systems, and geographically diverse ten-gigabit connections provide highly efficient data transfer throughout the country.
 

Figure 2. The company has facilities in New York, Lenaxa, and now Houston

 

Leasing an existing building eliminated the time required to build from the ground up, including time for obtaining a raft of construction permits. The warehouse, however, would not meet 1&1’s high standards for data-center efficiency, reliability, or security, especially since it’s located in the tornado-plagued Midwest. The solution: construct a freestanding building within the warehouse. The “building-within-a-building” design makes the facility itself redundant and protected from tornadic events.

Burr designed, built, and commissioned Phase I of the facility in six months. At 55,000 sq ft, the building provides more than enough space for offices and five data center rooms. Data center space totals about half the building’s square footage and is designed for 600 watts per sq ft for each room. The space provides capacity for years of aggressive growth and is highly energy efficient. The server rack system, for example, was specially designed and optimized before construction to run at 90 to 95 percent efficiency.

Additionally, 100 percent of the center’s energy consumption is offset by renewable energy through the purchase of Renewable Energy Certificates from the Bonneville Environmental Foundation. Globally, 1&1’s initiatives allow the company to avert the generation of 30,000 tons of CO2 annually.

The interior building’s 4-ft. raised floor is one of the facility’s most interesting features. It allowed chilled water piping and electrical services to be installed quickly, and avoided cutting into the existing concrete floor. Three completely independent and autonomous chilled water plants in a 2N+1 redundant configuration cool the facility. Each water plant has its own set of air-cooled chillers, pumps, and piping system for chilled water distribution. Thermal storage buffer tanks minimize the effect of a failed chiller on the system.

Computer room air conditioning (CRAC) units cool the server racks and pressurize the plenum below the raised floor. The CRACS in each server room are split evenly between the chilled water systems for added redundancy.
 

Figure 3. Exterior shots of the Lenaxa facility

 

During the design stage, Ziegler and lead electrical engineer, Doug Carder, ensured the data center would comply with the web-hosting company’s high operational standards. Automatic transfer switch technology, getting a second utility feed, meeting or exceeding electrical codes, and construction practices all received close scrutiny. It was essential that the power system have multiple stages of redundancy for seamless operation and reliability.

“Layered redundancy of power system equipment and chillers gives us flexibility in the way they can be used,” Ziegler said. “It had to be a sophisticated system.”

To help keep the project moving forward, Burr contacted ASCO Power Technologies to help with the fast-track timeline and for assistance with utility requirements concerning the use of closed transition transfer technology.

“Every data center is unique, and this one is no exception,” Chris McAnulty, the Southwest Region manager for ASCO Power Technologies.

Greg Sawyer, an electrical engineer with Burr, said, “We decided the best way to maintain redundancy was to have multiple methods of bypassing power distribution sections. This would allow system maintenance upstream and changing power distribution to the racks.”

The team selected, five 4,000-amp, closed-transition transfer switches to ensure power reliability. Four would power the loads and one would be redundant, and act as a swing switch to pick up any of the four other critical loads as needed.

The generators to power the system had a long lead time, which directly affected the fast track timeline. To hurdle this challenge, portable generators were installed until the stationary units arrived, which took more than a year.

Rather than parallel the 2250-kilowatt (kW) generators, 1&1 decided the redundant generator would be interconnected to all five electrical systems via high-speed static switches and make-before-break PLC-controlled (programmable logic controller) switch gear. This allows every generator, ATS, and switch gear to undergo maintenance one at a time, without dropping a load. Since all loads are a number one priority, the system arrangement allows for N+1 “plus” redundancy.

The gen sets are fed by fuel in a belly tank that will keep the data center operational for six hours. After that, supplemental tanks store enough fuel to keep the networking and computer systems operating for three days, respectively. Because 1&1 contracts with suppliers for fuel, the ability to maintain operation of the data center independently could be continued for weeks at a time if ever necessary.

To buy the precious seconds needed for the backup power system to start, Burr installed two 625–kilovolt-ampere (kVA) uninterruptable power systems (UPS). An upgrade to 750 kVA during phase-two construction makes certain that networking and computing equipment doesn’t suffer any down time and allows for additional growth. UPSs are redundant, as well.
 

Figure 3. Exterior shots of the Lenaxa facility

 

The entire system then underwent a battery of tests to ensure reliable operation. Ongoing monthly testing helps keep the system prepared for potential utility power outages. Since the data center doesn’t have a permanent load bank, gen sets carry the loads during testing. Breakers are tested quarterly on non-UPS loads.

In between testing, 1&1 monitors the system daily looking for operational trends. For security reasons, it’s only possible to monitor, not control, the on-site power system remotely. Operational control is possible only by direct contact with the equipment, which protects against potential negligent operation and hacking into the system.

Only six months after construction began, two of the five server rooms went operational with a third subsequently entering into service. All rooms have identical electrical systems to facilitate growth and system integration.

The data center’s redundant concept already has passed a real-life test. Last year, seasonal weather caused utility power to fail twice and the backup power system worked flawlessly. As an added measure of protection, an on-site weather station reports weather patterns that could possibly affect the data center and cause an outage. The engineering staff then fire up the gen-sets preemptively until the pattern clears.

Now, 1&1 executives can breathe a sigh of relief. They know the new data center can accommodate new customers and that future growth is covered.

“The only way we met our fast track timeline was everyone’s ability to understand the budget and delivery schedule,” Ziegler said. “Our reputation is based on reliable power, and it’s protected by the resiliency and flexibility of the power system design.”