The days of tossing equipment haphazardly into a temperature-controlled room and calling it a data center are over. To achieve maximum efficiency, reduce downtime, and minimize operational costs, today’s IT departments are focusing on a range of data center design best practices. It may take a bit more time and money to design a space that works, but the payoff can be significant. Consider the following tips as you lay out your next data center.

Standardize Your Cabinets & Racks

Although IT typically owns the data center, the equipment inside often belongs to or is controlled by various business units. Heterogeneity drives up costs, as it forces IT to uniquely manage each solution that’s brought in. Lack of standards compromises centralized management and impacts physical layout, which in turn leads to power, cooling, and floor space issues.

Jerry Allen, manager of Georgia State University’s Technology Operations Center (www.gsu.edu), says the data center can avoid compromise by imposing specific limits.

“When we’re asked to host equipment, if it doesn’t fit into our standard cabinet, then it probably won’t go into the data center,” he says. “It absolutely has to be rackable because we don’t have shelves to put workstation-type servers in here. It’s been difficult to get across to people that this is the way things have to be. But we’ve gotten there.”

If exceptions must be made, allocate at least one row for equipment that has unique space, power, or cooling needs. Avoid placing exceptional equipment in standardized rows or areas.

Keep Similar Technologies Together

Partitioning specific technologies into their own areas makes it easier to move them out—and move new equipment in—when systems reach end of life.

“A two-section layout of consistent technologies recognizes that you don’t install or remove all your equipment at exactly the same time,” says Ron Service, director of management information services for COM DEV (www.comdev.ca), a space hardware designer and manufacturer. "So when something new comes along, you can keep your unaffected equipment in place. This reduces labor, risk, and expense and maximizes uptime during any changeover period.”

At the same time, Service advises managers to watch floor density. “Don’t jam your data center to the point where you can’t function,” he says. “Getting new racks in and out as equipment expires should be easy in a well-laid-out room.”

Max Out Your Data Connections

It’s easy to underestimate the number of data connections required for a given piece of equipment. Georgia State University’s Allen says basic servers typically require three connections: foreground, background, and management. The numbers grow from there. More complex devices such as host servers may require up to eight network connections and four fiber connections. After installing 25 boxes into a 42U cabinet, Allen says things can quickly get out of hand.

“You need to be able to handle the growth of cable management,” he says. “Otherwise, you’ll have an unimaginable mess in back of your cabinets, and you’ll find it next to impossible to figure out which cable goes where.”

To leave sufficient room for future growth, Allen advises using wider 30-inch cabinets. “At minimum, we’d be using 24-inch-wide units,” he says. “The extra 6 inches of width in the 30-inch units lets us build cable management right into the rails. If you don’t, everything hangs out the back, you can’t close the doors, and you can’t do the things you need to do.”

Allen recommends against using patch panels. Instead, he uses switches to reduce the maximum length of any copper segment to 20 feet. Each switch manages up to three cabinets and is connected to a core router via fiber. The solution minimizes cable runs, improves airflow and cooling, and further eases cabinet access.

Create A Separate Test & Staging Area

Set up an area to work with new or nonstandard equipment before it goes into production, advises Allen, who recently did just that when some new SAN switches came in.

“We wanted to work with them before we felt ready to put them into production,” he says. “We needed enough room for fiber and power, and we didn’t want it mixed in with our production environment. But we still wanted it close enough for our people to conveniently work with.”

A staging area supports change management processes and avoids last-minute scrambles for power, cooling, and space.

Remote capability minimizes the need for systems administrators and other staff to spend time in the data center itself. Data center employees can provision equipment in the test area and then make it available remotely when it’s ready to be moved into production. Reduced physical traffic in the data center is always a plus, says Allen.

“The chance of an incident occurring is reduced by the number of hands in the cabinet or the number of times the door is opened,” he says. “Reduce the need for physical presence, and you improve reliability and reduce risk.”
 

Copyright ©2008 IT ACCESS. All rights reserved. Reproduction in whole or in part without permission is prohibited.

TO UNSUBSCRIBE
You have received this email because you are a newsletter subscriber. To unsubscribe from this newsletter, please
use the link below:
http://www.itaccess.com/news/unsubscribe.htm.

PRIVACY POLICY
Please read our Privacy Policy, or copy and paste this link into your browser:
http://www.itaccess.com/privacy.htm

For Further communication, please contact:
clientservices@itaccess.com
or
IT ACCESS
3 Sullivan Way
Westford, MA 01886
Phone: (978) 392-2245
Fax: (978) 268-5016