News

Data Center design: Form follows function

American architect Louis Sullivan, the father of modernism, is widely credited with the key axiom of 20th century modern architecture: “Form follows function.” That adage is also vital to data center design.

Sullivan put the axiom to work in 1891 in St. Louis with the design of one of the world's first skyscrapers, The Wainwright Building, and more than 110 years later MasterCard Worldwide employed Sullivan's axiom in the design of its data center in suburban St. Louis.

Indeed, when planning to update or construct a data center, functional requirements determine the design, old ideas mesh with new and technological advances feed a progressive approach. To maximize a data center build-out designers should examine three key concepts with “form follows function” axiom in mind: agility, reliability and cost.

Agility

While it may be peculiar to regard a mostly rigid, staid structure as agile, a data center requires agility to meet business demands. Agility in a data center is the ability to sense and react efficiently and effectively to environmental change. For better agility, a data center's form should follow its function:

* Make sure the raised floor and incorporated under-floor cable-tray system are of the right height and dimensions to facilitate organization and easy access for installation and equipment upgrades.

* Incorporate an equipment staging area separate from, but with proximity to, the data center. Shedding cartons, crates and packing materials outside the data center helps prevent contamination from dust and debris.

* When building a data center, design the floor to be level with adjacent floors, eliminating the need for entry ramps and making load-in and -out easier and safer.

* Construct a loading dock adjacent to the data center. A dock avoids the need for an external ramp, also making it easier to move large equipment in and out. Maximize ceiling height within the data center to facilitate heat dissipation and reduce server cooling costs.

* Incorporate hot/cold aisle controlled airflow layout, which can be essential in diminishing the risk of overheating and damaging equipment in enclosures and cabinets. By segregating cold air intakes (generally the front of equipment cabinets) from hot exhaust (typically expelled behind the cabinets), the direct transfer of hot exhaust air from one machine to the intake of another is nearly eliminated.

* Don't skimp on floor space. Plan for growth to meet data center demands for additional processing capability energy supplies, as well as increased storage capacity. And don't forget service requirements; ensure that your design allows for easy access by technicians as well as equipment movement. By outlining your anticipated requirements and the footprint of the equipment on the floor, you'll avoid overlooking important requirements.

Reliability

Form and function are important components of data center reliability. The 2N+1 concept is an easy calculation for commodities such as energy, capacity and storage; consider the extrapolation for the physical plant:

* In building its St. Louis data center, MasterCard designed it to the seismic standards of earthquake-prone California, and constructed it to withstand sustained mph winds. Its electrical system is built to near-disaster-tolerant specs.

* Multiple, diversely routed energy sources are another uptime ticket. The MasterCard data center has access to two commercial electrical stations and is backed up by prime-duty generators and a cache of fuel – both always at the ready.

* Don't let your data center be threatened by something as mundane as utility dredging on the perimeter of the property. Consider the robustness of underground conduit, and consider how the growth of equipment and cabling inside the data center walls will impact the surrounding ductwork.

Cost-effectiveness

* Obviously monetary resources are not limitless, so cost-effective design is fundamental to long-term success. For better cost-effectiveness, again, form should follow function:

* Data center managers should demand a voice and fully participate in all new data center strategy, design, construction and budget meetings. Accountants and facilities personnel not CTOs. Eliminating or limiting features or budget-line items may seem effective on paper, but ill-considered cuts could result in serious service interruptions downstream.

* As much as possible, plan for the future. It's much easier and ultimately more cost effective to implement needed features in build-out than to retrofit. Pay close attention to data center and equipment innovations to determine forward-looking budget strategy. For instance, cooling has come full circle. Previously, mainframes were water cooled, followed by servers cooled by air. Now, many companies are advocating solely water cooling with integrated water-rack cooling systems. Racks alone are a prime example of changing technology, increasing in sophistication over the years. So, if you are building a data center today, consider the need for incorporating water cooling.

* More companies are going green because customers are demanding it – they want to do business with companies that are environmentally responsible. Beyond that, there may be inherent cost savings; for instance, design the data center roof to collect rain water for reuse in server cooling.

* A growing number of companies are conforming to Leadership in Energy and Environmental Design certification. LEED recognizes performance in five key areas of human and environmental health: sustainable site development, water savings, energy efficiency, materials selection and indoor environmental quality. The LEED Green Building Rating System encourages and accelerates global adoption of sustainable green building and development practices through the creation and implementation of universally understood and accepted tools and performance criteria. It is a third-party certification program and the nationally accepted benchmark for the design, construction and operation of high-performance green buildings. LEED certification speaks volumes about your commitment to a green future.

* Develop partnerships with key vendors. Buy from multiple providers to leverage competition, but also evaluate strategic vendor partnerships. Consider adopting the philosophy that even partners need to earn your business each and every day.

* Adopt virtualization technology, using software to scale out rather than up. This reduces server footprint and uses a greater percentage of the processing capacity that you already own. Through virtualization in the MasterCard data center, in the Windows world, the number of server instances is nearly double the actual server population.

These best practices coalesce into a cohesive data center approach that Sullivan likely would endorse: Focus on the future with an eye toward the wisdom of the past. It may seem rudimentary to ensure a 21st century data center is modern, but often the greatest wisdom is found in the simplest ideas. When form follows function in agility, reliability and cost, what follows is data center success.

Previous ArticleNext Article

Leave a Reply

GET TAHAWUL TECH IN YOUR INBOX

The free newsletter covering the top industry headlines

Send this to a friend