• Simplicity in identifying and implementing energy efficiency measures can capture low-hanging fruit that delivers the most bang for the buck.
• Establishing a baseline can help you identify whether an energy efficiency strategy is actually improving efficiency and providing cost savings.
• Overcapacity is the enemy of cost-effective energy management and consumption. Administrators should understand their infrastructure and their needs and then plan accordingly.
As computing density in data centers increases, the focus on capturing savings by implementing energy efficiency measures should increase accordingly. The math is simple: The more equipment in a data center, the more energy is consumed. And energy costs, if left unchecked, can take a substantial bite out of a budget.
But before diving in, administrators should think carefully about the path forward for conducting energy efficiency audits and implementing energy efficiency measures. Mistakes here can result in inefficient measures that don’t deliver results and wind up costing valuable time and money. Read on to learn more about these potential mistakes and what to do to avoid them.
Keep It Simple
David Matusow, Center of Excellence architect at Sungard Availability Services (www.availability.sungard.com), says one of the biggest mistakes is looking for the exotic. Often, he adds, it’s the simple solutions that are both the easiest to implement and the fastest to get done and have very significant areas of improvement.
For example, Matusow says simple things such as implementing sleep modes on servers and lighting can result in very sizeable gains without the need to incur capital expenditures requiring top management approval levels.
Matusow says Verizon did a study in 2008 on the sleep feature on most PCs; the study showed the company would save an estimated $7 million per year just by implementing this feature at night when PCs were not utilized. According to the study, Verizon could achieve energy savings equivalent to that needed to light nearly 88,000 homes annually.
But there is a caveat, says Matusow: It is very common to look for savings, root out inefficiencies, and ensure an environment is as green as possible only to forget that the business needs to be organizationally and financially responsible. It is important to assess the larger picture to determine where the lines are drawn, he adds. Administrators should ask themselves at what point is it worthwhile to spend a million to save a thousand.
Also, he adds, organizations often look at the immediate and direct benefits of energy efficiency measures without considering the indirect and peripheral benefits. For example, if a data center’s energy consumption is a burden on the power infrastructure and causes a direct negative effect on community stakeholders, it might be appropriate to implement energy efficiency measures that reduce the impact.
Neil Kimberly, solutions engineer at 1E (www.1e.com), says the most common mistake IT managers make when implementing power management measures is failing to create a baseline to measure against each step of the way to ensure a power management project leads to real savings.
Sungard’s Matusow emphasizes that it is important to get intimately acquainted with the data center. A fact-finding mission to identify what is being spent and where costs are originating from should involve discussions with the data center personnel who are knowledgeable about their environments and know how to best address efficiency, he adds.
Only when the full assessment and fact finding are done should administrators pursue solutions. One cannot find facts, Matusow emphasizes, without knowing the context; these fact-finding efforts often uncover locations for improvements in areas that one would have never expected.
Keep It Realistic
A common mistake that can lead to substantial energy overages is the tendency by data center administrators to install excess capacity, be it cooling, servers, or other items. It is important to accurately determine needs based on real operating conditions and plan purchases accordingly.
For example, a potential land mine is the failure to properly determine the cooling requirements at the data center, says Eric Hansel, president of EGM Green (www.egmgreen.com). Hansel recommends that administrators figure out exactly how much cooling will be required. Once that’s done, they should install a high-efficiency cooling unit with the correct tonnage. Other areas where savings might be captured are in lighting within the data center and the use of high-efficiency processors and servers.
Another mistake along these lines, says Daniel Feldman, director of telecom marketing at Microsemi (www.microsemi.com), is having PoE (Power over Ethernet) at every port. Feldman says that PoE ports, whether they are turned on or off, mean the system power supply is oversized, so there is higher consumption even when idle.
According to Feldman, ways to overcome this include using midspans where possible, scheduling ports that are not needed at certain times to completely shut down, and using equipment that relies on PoE for power because such equipment tends to be more efficient than its non-PoE counterparts.
by Sixto Ortiz Jr.
Best Lesson Learned: Cost Analysis Is Key |
A simple—but potentially costly—mistake administrators can make in terms of auditing or implementing energy efficiency measures in the data center is doing nothing at all. Unfortunately, says Tyson Hartman, global chief technology officer at Avanade (www.avanade.com), more than half of the respondents to a recent survey commissioned by Avanade said that even though energy is a top cost, these companies still fail to account for energy costs when developing IT budgets.
The study shows that even though more than 50% of all companies surveyed said they measure all costs associated with a system before implementing and 89% cited cost as a significant factor in a company buying decision, only 25% list energy among their top three largest IT costs per year.
Hartman points to a Gartner estimate, which, alarmingly, states that over the next few years, most enterprise data centers will spend as much on energy as they do on hardware infrastructure. Hartman says companies need to begin considering greener options such as virtualization and cloud computing so they can adjust their IT budgets appropriately to curb these rising costs.