|
 |
|
General Information
|
Add To My Personal Library |
January 14, 2010
• Vol.32 Issue 2
Page(s) 1 in print issue
|
Harness Heat & Power Consumption
Get Your Data Center Out Of The Hot Seat In 2010
|
| Key Points
• Targeting temperature-related problems requires close analysis of not only your current environment but also your existing systems and their dependencies on other parts of the data center.
• By working to cut power consumption, you can help to control temperatures while saving on perpetually rising energy costs.
• Depending on the scope of the project, costs can range from a few hundred dollars to tens of thousands, but keeping an eye toward reducing energy consumption can help to offset initial investments. | | Data centers aren’t the only things chugging along into the new year, as heat is anticipated to have yet another successful year of delivering headaches to infrastructure managers around the globe. However, by undertaking a project in 2010 that can help to keep temperatures and power consumption under control, data centers can force heat to look elsewhere to continue its productive run.
An easy approach to controlling rising temperatures is simply to throw more cooling at the heat, but then the data center faces increased power consumption costs. The challenge, then, is to achieve acceptable temperature levels while keeping an active eye on green IT measures. This process is by no means simple, but it’s also not necessarily expensive nor out of the expertise realm of even small data centers.
 Hit The Heat
Analysis is key to understanding the temperature range that your data center should achieve both now and for the next several years, says Rudy Rangel, sales manager for Rackmount Solutions (www.rackmountsolutions.net). He recommends using temperature monitors to visualize current temperatures and see how they fluctuate during the day and various seasons. From there, ask questions: Are there hot spots? Are the temperatures on nights and weekends set higher than during weekdays? Are you determining your ideal range based on typical server-room wisdom of 68 to 72 degrees Fahrenheit?
 Also, determine whether you are having equipment failure now or if you can safely boost temperatures. For example, Rangel notes that some manufacturers now recommend operating temperatures of up to 82 F, but it remains a good idea to check with your own equipment manufacturers before assuming that your data center is overheated. It’s also wise to understand how that equipment interacts with the rest of the data center.
“The first step is to discover the IT infrastructure and facilities,” says Dr. Mickey S. Zandi, managing principal, consulting services, at Sungard Availability Services (www.availability.sungard.com). “You must understand the current state of the infrastructure—how is it interconnected and independent? By understanding the interdependencies, an organization is then able to right-size the infrastructure. Next, validate your design layout and modify it to meet current trends. Carefully plan the change to the topology and the interconnectivity.”
According to Rangel, if your data center’s temperature needs to be adjusted only a few degrees, or if it experiences constantly fluctuating temperatures, a series of small fixes could satisfy the project. For example, control in/out access to the room to harness unnecessary temperature fluctuations, and implement spot coolers directed at hot spots or to boost the current air-conditioning system. He also recommends installing one or two dedicated AC cabinets for the hottest configurations, as well as adding ducts and vents to the server room’s existing AC system.
Power Down
Keeping temperatures under control is critical for the efficient operation of your data center, but undertaking this project without regard to power consumption can lead to escalating costs. So, while working toward temperature management, be sure to watch for opportunities to control costs through green infrastructure and power management systems.
“Not only does a sustainability, or green IT initiative, present a substantial potential for cost savings in the face of escalating power costs, it offers the data center to present its ‘green’ credentials to customers, co-workers, and affiliates,” says Steve Kolbe, CEO and founder of AnalySys (www.analysys.net). “It also is a proactive approach to a likely eventuality that the U.S. data center may become regulated as cap-and-trade or other similar legislation such as the U.K.’s Climate Change Levy [a carbon tax] is enacted to stem the further growth in consumption.”
Kolbe recommends taking inventory of systems presently in place and reducing in-production systems where possible. For example, you can virtualize servers and identify systems that can be eliminated or transitioned into a powered-off, or cold backup, state. Not only will these steps help to keep energy consumption under control, but they’ll also contribute to the overall effort of keeping temperatures in check. Kolbe also suggests identifying systems that can be targeted for scheduled or scripted sleep mode or power-off status.
Required Resources
Attacking rising temperatures while keeping power consumption in check can easily empty a yearlong budget, but that’s not a given. In fact, some measures can help to save tremendous amounts of money over time, thanks to a better-running data center infrastructure that isn’t handcuffed by pricey energy costs.
Rangel says that simple cooling fixes can range from several hundred dollars to about $20,000 if you purchase several dedicated AC server cabinets or a new AC unit. However, if you currently have no raised flooring but determine that you need it, you could be facing costs of up to $100,000 for architectural fees, builder’s fees, and the product itself. But solutions geared toward reducing energy consumption offer a more cost-neutral proposition, because the savings on power costs can eclipse the one-time cost of systems themselves.
In terms of personnel, this project relies on several parties to help reach a successful conclusion. Rangel notes that IT staff members are required to reroute cables and rerack equipment as needed, and facilities management and staff are needed to move equipment, add AC ducts, and place tighter seals and/or controlled access locks on entry and exit doors. If new equipment is required for the project, vendors will also be in the mix. Finally, Kolbe recommends involving an outside electrician for low- and high-voltage wiring as well as a consultant qualified in green or sustainable technology initiatives. 
by Christian Perry
Where To Start
Big infrastructure projects should never be taken lightly, particularly because you’re dealing with the backbone of your data center. Steve Kolbe, CEO and founder of Analysys (www.analysys.net), says that it’s a good idea to take a slow, deliberate approach to the project.
“Data centers are expected to function seamlessly 24/7, without outage or incident,” he says. “Therefore, a proof-of-concept step is appropriate on nonproduction systems. This allows the implementer to demonstrate functionality, and it also gives insight into the reality of any anticipated cost savings that may have been calculated, or proposed if an outside consultant is involved.”
Before jumping into the proof of concept, Kolbe says it’s necessary to conduct comprehensive analysis, which can take at least two months to establish an appropriate average. This analysis should depend on a strategy that looks at current inventory and determines where changes can and should be made. |
|
|
