The traditional priority in data center management has been IT systems availability and uptime. Energy costs were a secondary consideration managed after the top priorities were seen to, often resulting in data centers being overbuilt and underutilized.
But today, with rising energy costs and more companies doing carbon accounting, as well as changes wrought by “denser” computing due to technologies such as virtualization, energy management has become a top priority for many data centers.
To avoid energy waste, a basic goal with data centers is to ensure that as much energy as possible is consumed by the IT gear rather than infrastructure such as cooling units, fans, and uninterruptable power supply (UPS) equipment. A key metric in data center energy management—the Power Usage Effectiveness (PUE) metric—tracks this proportion of total data center input power to IT load power.
As detailed in white paper 154, “Electrical Efficiency Measurements for Data Centers,” a PUE of 1 would indicate perfect efficiency—with all power reaching the IT loads. Most data centers are no where near this—with a PUE of 1.25 to 1.3 considered exemplary.
Of course, data center energy management involves much more than using one metric. For starters, PUE should be used in the context of an efficiency model for improvement, which might involve audits and other services to establish. There also are a variety of products that are key enablers of energy management. The good news is potential for improvement. As white paper 161 explains, the energy savings opportunity from managing the physical infrastructure in a typical data center ranges from 10 percent to 40 percent.
There are many solutions that improve energy efficiency. For example, newer UPS systems are significantly more efficient than the typical older, installed based of UPS units, especially at lighter loads. As explained in white paper 126, one study found that at 30 percent load, newer UPS systems pick up more than a 10 percent efficiency gain versus the average of installed UPS.
When it comes to energy management, one of the most important solutions is data center infrastructure management (DCIM) software. DCIM suites span functions including monitoring, alerting, capacity planning, as well as energy efficiency tracking.
The power and cooling equipment used in many newer data centers tends to be more modular and scalable, offering greater potential for energy efficiencies and savings. For example, modular “in row” cooling puts the cooling closer to the source of heat at the computing load, versus the more traditional computer room air conditioning units which cool the data center from the perimeter. When combined with containment barriers to segregate hot air, in row-cooling can be highly efficient, and also well suited to the rapidly shifting computing loads seen with server virtualization. Today, best-in-class components for cooling, containment, and power distribution can be fitted together quickly to support a “pay-as-you grow” strategy that avoids overbuilt and underutilized assets.
Services play a crucial role in assessing which solutions will improve on data center efficiency, and in helping companies establish a better foundation for energy management and planning. These services include:
- On-site evaluation and assessment of data centers, such as energy assessment services that spot inefficiencies. Sometimes, gains can be made via minor adjustments revealed by audits, such as filters that aren’t being changed, or poorly placed air vents.
- DCIM implementation services and post deployment audits.
- Outsourced services, such as remote monitoring and diagnostics of data center critical infrastructure.
While much can be done to improve energy efficiency in existing data centers, energy efficiency should be considered during the design phase for data centers or major retrofits. By taking into account approaches such as modular, scalable cooling, new facilities are “right-sized” from start, and can scale as needs grow.
In sum, energy management encompasses many technologies and services. It works best under a life-cycle approach to all the factors that influence efficiently getting energy to the IT load and heat removed from the data center.