News

Technologies & Processes To Keep In Mind For A Well-Organized Data Center

The road to a well-organized, effectively designed data center begins during the design and development phases. It is often the case that unexpected—and unwanted—data center events occur due to flaws introduced during the data center design phase.

Fortunately, mitigating a lot of these issues is often a matter of thinking about potential problems during the data center setup and layout phase and designing around them. Doing so will ensure that a new data center is designed for minimal downtime as well as effective use of space and infrastructure elements.

Consider Virtualization & Cloud Options

Virtualization is rapidly becoming a commonplace technology in use in many data centers. That should come as no surprise: Virtualization can greatly reduce server sprawl while maintaining the processing capabilities business applications need to work most effectively. Minimizing server sprawl can greatly contribute to a well-organized data center simply by reducing the amount of hardware that needs to be hosted.

Bradley Brodkin, president of HighVail Systems (www.highvail.com), says a good first step during the data center setup and layout process is for administrators to look at all existing applications, services, and processes currently running in the data center to determine the amount of application load that can be consolidated on a virtualized server, farm, or private cloud to achieve higher resource utilization.

However, virtualization is not a cure for all ills. Jon Heimerl, director of strategic security at Solutionary (www.solutionary.com), says administrators should ensure that they balance applications that have different demand cycles on the same hardware and stay away from, for example, packing a single server with 60 virtual systems that all have their highest demand cycle between 12 and 2 p.m. on weekdays. It’s important to understand that virtualization can potentially increase overall system utilization and power use, Heimerl says.

Along similar lines, Brodkin says a public/private cloud can provide flexibility by allowing administrators to plan for balanced loads and farm out computing power to meet infrequent peaks and valleys in requirements.

Implement Cable Organization

Cabling may seem like a pedestrian task, but effective, well-thought-out cabling design can mitigate many headaches during the life of a data center. Dealing with a rat’s nest of tangled cables is not a good use of time, especially when the clock is ticking and a data center issue needs to be promptly resolved.

Michael Frank, vice president of data center services at Internap (www.internap.com), says properly sized cabling makes it easier to troubleshoot any challenges that may arise in interconnected equipment. During the design phase, admins should avoid landing on deployments that result in confusing tangles of wires that cannot be easily traced when equipment problems occur. Also, designing tight turns and bad bends can degrade signal quality and limit throughput.

Purchasing the correct power strips allows for power cables to be neatly routed and managed. For example, says Frank, the use of vertical and horizontal power strips is dependent on the overall design and the cabinets or racks into which equipment can be deployed. Finally, Frank says, administrators should develop a network plan that anticipates growth so cable runs and connection points can be minimized. Shorter cable runs and fewer connection points improve troubleshooting and minimize points of failure, he says.

Heimerl says admins should include support for additional equipment in their cable capacity planning, such as environmental sensors, tamper indicators, and surveillance camera needs. For more information on smart cable management, see “Cable Organization & Management” on page 28.

Plan For The Unexpected & Know Your Building

Heimerl says data center design should include planning for unexpected events. For example, he says, even though a raised floor may not be needed because everything is supplied from above, administrators should not ignore the fact that the raised floor provides another function. Heimerl tells the story of a data center located in a building basement that did not have a raised floor. When a fountain at the front of the building leaked, 3 inches of water flowed into the data center, causing rack power supplies located close to the floor to short out. And, he adds, the data center had no floor drain, so water had to be manually bailed out.

Heimerl also advises admins to maintain an accurate “as-built” blueprint of the data center that clearly shows where key infrastructure items such as utilities and cabling are located. Heimerl cites an engineer who worked on an upgrade at a data center and needed to cut a hole in a wall to run new conduit. The engineer did not have drawings available, so he cut the hole in a spot on the wall that he felt was suitable for the task. Unfortunately, the engineer cut into the sewer line for the five-story building where the data center is housed. The hole in the line promptly drained 6 inches of sewage into the data center. Sadly, the data center also lacked a floor drain.

Pay Attention To Cabinets & Racks

Internap’s Frank says facility administrators should procure properly sized cabinets and racks with adjustable mounting rails to allow for the placement of equipment both today and in the future with less chances for complications as IT equipment changes.

Also, he adds, IT administrators should understand the cabinets or racks available in the data center because these will sometimes dictate what equipment will work and what won’t, as well as any special equipment required, such as shelves or sliding rails. Administrators should also watch the cabinet depth and width, especially if vertical mounted power strips, wire managers, and relatively deep servers are used.