Revamping Existing Centers Is A Potentially Budget-Friendly & Effective Trend To Watch
Key Points
• With the implementation of more blade servers, many data centers are opting to redesign rather than to use managed services or undertake a new construction.
• When taking on a redesign, first quantify the overall capacity of the data center in terms of space, power, and cooling.
• Run the numbers to make sure that redesign makes sense; for some data centers, the cost of new construction might actually be lower than taking on demolition and rebuilding.
Because building a new data center can be capital-intensive, many companies are choosing to redesign existing centers, particularly to address power density issues. Although this type of revamp may require careful planning, many experts believe it’s worth the extra effort and expense to stay up-to-date with newer technologies in such areas as cooling and power.
Growing Trend
Even in a soft economy, data center design companies are still quite busy, notes Peter Sacco, president of PTS Data Center Solutions (www.ptsdcs.com). “I think that can be attributed, in part, to that fact that when companies do layoffs, they have to do more with the people they retain,” he says. “And for that, you have to automate processes, and that leads to implementing technology to see those productivity gains.”
About 70% of his clients are using blade servers, he adds, as well as upgrading SANs, but many haven’t revamped their facilities to accommodate the changes. Sacco believes this is typical in the industry and that many companies are facing the challenges of dealing with higher-density equipment—leading to power and cooling issues—and are leaning toward using their own data center spaces rather than colocation or managed services.
In general, most redesigns today are incorporating the ability to handle higher power densities, dissipate heat more efficiently, and utilize greener building standards, notes Kris Domich, principal consultant at Dimension Data (www.dimensiondata.com).
“With additional power comes additional heat, so redesigns that involve the delivery of more power must take into account how they will evacuate the heat,” he says. “The trend now is not only to add a sufficient amount of CRAC/CRAH devices but to also engineer the physical space to provide dedicated paths for the removal of hot air.”
Toward that end, one trend has been the introduction of more generators and UPS systems to the data center, says Ron Hughes, president of California Data Center Design Group (www.cdcdg.com).
“To address power density and cooling, people are looking at implementing supplemental cooling systems, which are sometimes easier to put into a retrofit,” he says, adding that the biggest challenge tends to be lack of floor space for PDUs but that many centers are gaining efficiency by using in-row systems instead.
Another consideration is use of outside air, adds Hughes. Particularly for data centers in cooler parts of the country, piping in cold air can create dramatic savings in cooling costs. “We’re going to see the use of nontraditional cooling increase as a trend because we’re not talking about savings of 1 to 2% [but] more like 30 to 40% with some of these strategies,” he says.
Redesign Issues
When considering a redesign, the first step is to quantify the overall capacity of the data center from space, power, and cooling perspectives, advises Domich. Each of these elements makes up a finite “unit of currency” across the data center, and running out of any of these elements can end the data center’s ability to expand.
Next, map the consumption of these elements back to physical resources—network, storage, servers, mechanical, and electrical—and attempt to understand the anticipated growth of each group of resources.
“It’s important to understand the growth rates of each resource type because each one will consume space, power, and cooling differently,” Domich says. “By tracking these growth rates against available capacity, it should become clear as to when you run out of overall capacity.”
Gaining this visibility will help answer one fundamental question, he adds: Is there even time to redesign? “A redesign is only necessary if by doing so you stand to recapture stranded resources or if the redesign itself will increase the overall capacity of space, power, and cooling, which is rare unless the physical space is being enlarged as part of the redesign,” Domich says.
Cautionary Note
Even though redesigning is a strong trend, IT managers do need to make sure that the costs make sense, notes Domich. He says that depending on the conditions of the existing center, the total cost of demolition and rebuilding can easily rival the cost of building an entirely new data center. A major consideration is the cost of temporary housing of existing systems during the construction period.
“The main thing we stress to clients is the total cost of ownership of the data center and to ask themselves if running a data center is a core competency,” says Domich. “If not, organizations should explore the colocation option.”
For those that do choose to tackle redesign, it’s likely that efficiency benefits can be reaped if a long-term view is taken. As Domich points out, a data center should be revamped to last at least three years after a reorganization. With a longer timeframe like that, IT managers will have more options when it comes to tackling density issues.
Article by Elizabeth Millard for 
.