Pinching Pennies Doesn’t Mean Cutting Back On Efficiency

Key Points

• Cutting power costs is a two-part problem: knowing what is happening at a power level down to the circuit and identifying inefficient pieces of equipment.
• Increasing the supply set points for CRAC (computer room air conditioner) units can have a huge impact on cooling costs.
• Small efforts such as monitoring usage, sealing airflow, and keeping equipment up-to-date can add up to big savings.

When CFOs and CEOs tell IT and data center managers to trim the excess off their budgets, power and cooling are two of the usual targets. Although they are both integral players in data center efficiency, they can’t hide forever when cost cutting is the name of the game. Here are some tips and advice to use when you need to do more with less with power and cooling.

Cut Down On Clutter

De-cluttering the data center is something that any organization should consider. “What is in the room that does not absolutely need to be in the room?” asks Melanie Davy, research analyst for Info-Tech. “For example, fax servers or equipment that does not require an extra layer of redundancy can be moved out of the data center and into the wiring closet.”

Davy says the energy savings here lies in standby power, where equipment that does not require a constant supply (and is not mission-critical) no longer consumes data center power.

Efficiency Is Key

For Stephan Prueger, vice president of sales with TrendPoint Systems (www.trendpoint.com), it’s all about focusing on efficiency. “We’re talking about the efficiency of IT equipment, the cooling scheme, and the AC equipment,” Prueger says, adding that cutting costs in power is really a two-part problem: knowing your power usage at a granular level and identifying which equipment is inefficient. “You need to know what is happening at a power level down to the circuit. Circuit granularity enables you to monitor not only the racks but all the devices in the data center, including the air conditioning,” Prueger says. “Identifying where most of the wattage is going identifies the hot spots and allows you to balance the load and match the cooling. Now when you have power factored at the circuit level, you can identify inefficient pieces of equipment.” This lets you narrow down what equipment is inefficient and replace or tune that equipment, he says.

David Matusow, a Centre of Excellence architect at SunGard Availability Services (www.availability.sungard.com), says companies must start to review their energy demands now and not wait until a hard limit is reached. “Reviewing now provides a company with the time and insight required to make adjustments prior to hitting a solid stop point,” he says. “If a power limit is reached, a company must make difficult choices—are data center plans postponed or priorities? What is needed is an effective way to measure the current state and project impacts and demands.” Matusow says capacity planning must include not only processor cycles but also infrastructure requirements.

Small Changes, Big Savings

Rick Berendes, co-founder and product manager at Data Center Resources (www.datacenterresources.com), says when it comes to cutting cooling costs, increasing the supply set points for CRAC (computer room air conditioner) units can have a huge impact on cooling costs. “At higher set points, cooling units do not run as hard or as often,” he says. “The trick is to be able to do this while maintaining suitable conditions for servers. That is the reason we are seeing data centers do whatever they can to make their existing cooling infrastructure as effective as possible.”

According to Berendes, common practices include sealing cable openings, using blanking panels in unused rack spaces, and directing airflow in supply and return plenums and aisle containment. “These approaches can help managers get the most out of their cooling capacity, which may allow them to start to increase set points,” he adds. “However, it is important to have a real-time temperature monitoring solution in place, as well. This will allow a site to incrementally ratchet up supply temperatures while ensuring that servers are not in jeopardy along the way.”

Stay Up-To-Date

Cooling the data center can amount to about 50% of the operating cost in the data center, according to Davy. She says most organizations can lower energy costs in the data center simply by refreshing their equipment, because newer cooling infrastructure is more efficient than older equipment. “For example,” Davy notes, “the average life span of a data center is between seven and 12 years. During that time, equipment and technology will change two to three times (cooling and other IT equipment). A typical commercial CRAC unit used in data centers 10 to 15 years ago would have a maximum capacity of 7 watts per square foot.”

However, according to Davy, basic CRACs today operate at around 50 watts per square foot, and some higher-end precision and balanced solutions can achieve up to 250 watts per square foot. She adds, “Today’s air exchange precision systems pass more than 500 cubic feet per minute per ton, while previous technologies might pass as little as 350 cubic feet per minute per ton.”

Prueger says when you match the cooling to the wattage (directly proportional to heat), you more efficiently cool your equipment. “Take the dynamic load from the power monitoring in your data center and match it to the amount of cooling you are pumping into your data center to efficiently maintain a steady temperature,” he says, adding that more efficient systems result in lower PUE (power usage effectiveness).

Article by Chris A. MacKinnon for .