|
 |
|
General Information
|
Add To My Personal Library |
September 11, 2009
• Vol.31 Issue 23
Page(s) 42 in print issue
|
Optimize Data Center Cooling
Keep Your Cooling In Tip-Top Shape To Help Conserve Energy & Money
|
The science of keeping a data center cool is an essential piece of the data center operations puzzle, and until processor technology makes a radical step toward the development of more heat- and energy-efficient designs, the problem of heat generation and dissipation will continue to dog administrators. A poorly cooled data center can become the proverbial anchor around an organization’s neck, giving rise to unpredictable, spotty performance and lost revenues and productivity.
Fortunately, there are steps administrators can take to optimize cooling in the data center. Read on to learn how to make the most of your cooling infrastructure.
 Measure What You Have
Before any effective changes can be made, administrators must understand what they have to work with. Without that knowledge, there is no way to determine how effective a proposed cooling optimization solution might be and thus the return on investment it is able to deliver.
“If you are not already capturing detailed information on existing conditions and calculating your site’s power usage effectiveness (PUE), it is crucial to document conditions such as utility load supporting the data center, IT equipment intake temperatures, UPS load, [and] redundant cooling capacity before implementing any changes,” says Lars Strong, an engineer at Upsite Technologies (www.upsitetechnologies.com). This will enable administrators to document improvements and communicate results to upper management, he adds.
“If you can’t measure it, it’s hard to manage it,” says David Brown, president of Datotel (www.datotel.com), a St. Louis-based colocation and outsourcing facility. Brown recommends that administrators either automatically or manually monitor the temperature at various heights in the cabinets. Also, he adds, administrators should ensure that temperature ranges are maintained within predetermined service levels.
Get A Data Center Health Check-Up
If comprehensive monitoring and PUE calculations are not part of data center management practices, says Strong, a computer room cooling efficiency health check can help administrators identify how improvements to the airflow and cooling dynamics of a computer room can reduce operating costs, increase capacity, and defer capital costs.
Strong recommends that a checkup of cooling system effectiveness should involve an examination of three aspects of data center health: the condition of IT equipment air intakes measured in temperature and humidity, the percentage of bypass airflow, and the CCF (cooling capacity factor).
Strong defines the CCF as the margin of installed cooling capacity vs. the load. A number of measurements need to be taken in order to determine the CCF, including the measurement of cabinet air intake temperatures, relative humidity of any identified hot spots, maximum temperature drops across cooling units, and delivered cooling unit airflow volumes. In addition, the delivered cooling capacity of each cooling unit should be calculated and compared to manufacturer specs, the computing equipment power load should be summed up, and instances of latent cooling and latent cooling penalties should be determined. Finally, all return air temperature and relative humidity sensors should be checked for proper calibration.
“These measurements and resulting calculations will provide clear indication of opportunities for improvement,” says Strong.
Plug Leaks
Cool air leakage is a significant obstacle to the optimization of data center cooling. Significant leakage results in massive wasted energy and, in extreme cases, could lead to equipment downtime.
Jack Tison, vice president of technology at Panduit (www.panduit.com), points out that a significant amount of cool air can pass through improperly sealed spaces within racks, cabinets, and floor cutouts. This situation disrupts the efficient movement of air, reduces the efficiency of equipment blower fans, and wastes energy used to cool the air.
Tison says administrators can remedy these leaks by placing fillers or blanking panels in empty rack spaces to promote cool airflow through server and switch air intakes. Also, he adds, air-sealing grommets should be used to seal cable cutout spaces under cabinets, eliminating bypass air and increasing the static pressure under the raised floor.
Joe Perillo, vice president of enterprise infrastructure technology operations at MetLife, says his organization uses a multipronged approach to ensure efficient airflow in the data center: Server racks are arranged in a hot aisle/cold aisle configuration, with perforated tiles positioned on the cold aisle floor and hot aisle ceiling; hot-air removal units are attached to the server racks; and grommets are used to minimize bypass airflow through the floor. Finally, says Perillo, cables are maintained above the raised floor—if the rack configuration allows it—to optimize airflow below the raised floor.
Other places to look for leaks, says Upsite Technologies’ Strong, include column façades to adjacent floors, entrance doors and elevators, loading dock doors, windows, overhead wall openings where cables pass through, and holes in the perimeter walls above the dropped ceiling. In addition, all openings in the raised floor that do not deliver conditioned airflow directly to the intakes of IT equipment should be sealed. These include cable openings under or behind cabinets and holes under PDUs or for conduit penetrations.
Take Advantage Of Free Cooling
There is no quicker way to capture energy savings than to use the cooling capabilities that Mother Nature already provides. MetLife’s Perillo points out that depending on the location of the data center, administrators may be able to take advantage of “free cooling.” For example, says Perillo, MetLife takes advantage of cold winters by using water-side economizers, which use cooling towers to cool water using low outside temperatures.
Given adequate ambient environment, a water-side economizer eliminates the need for cooling with chillers, which is advantageous because chillers are one of the highest energy-consuming devices in a commercial building. Panduit’s Tison says air-side economizers can also provide savings by bringing outside cool air in, filtering it, and then expelling waste heat back to the outside. 
by Sixto Ortiz Jr.
BONUS TIPS
Avoid cable clutter. According to Jack Tison, vice president of technology at Panduit (www.panduit.com), cable management best practices such as routing data and power cables toward cabinet side walls away from server exhaust areas and properly placing cable pathways outside the cabinet can reduce clutter and eliminate exhaust area choke points, improving cooling efficiency.
Speak to local electric utility companies. David Brown, president of Datotel (www.datotel.com), says an increasing number of electric utilities offer programs such as free consulting, audits, or economic incentives to help data centers become more efficient and use less power. |
Easiest Tip: Keep Up With Preventive Maintenance
David Brown, president of Datotel (www.datotel.com), says preventive maintenance is an obvious area where administrators can capture cooling system optimization gains.
Brown says that poorly maintained equipment not only increases the likelihood of failure in the cooling infrastructure—which can cause heat issues all by itself—but is also more costly to operate and won’t run at full cooling capacity. So, he emphasizes, administrators should stay on top of their cooling equipment maintenance. |
Best Return On Investment: Consolidate With Virtualization
At the end of the day, cooling optimization is required in order to keep data center equipment such as servers and networking gear operating most effectively. Therefore, an excellent way to optimize data center cooling is to reduce the number of servers in the data center. In this area, virtualization can help.
Joe Perillo, vice president of enterprise infrastructure technology operations at MetLife, says his company has achieved significant cost avoidance since it started using server and storage virtualization in 2006. Despite the ever-increasing demand for computing power and storage capacity, he says, the company has been able to decrease energy consumption in power and cooling by reducing the physical number of servers needing to be cooled.
|
|
|
