Data Centers Don't Have to Choose between Uptime and Efficiency

A 2014 report from the International Energy Agency concluded that energy efficiency investments could potentially create total returns of $18 trillion worldwide. This staggering number – which is roughly the equivalent of all of North America's economies combined – truly shows the value in adopting energy efficient measures across the consumer and business landscape.

Taking a closer look at where and how our power is being consumed can help determine where energy efficiency efforts should be focused. The US Department of Energy found that data centers account for about 2 percent of our nation’s electricity use, and the rate is climbing – making data centers a prime target for implementing energy efficiency measures.

The US Government agrees. The Better Buildings Challenge, sponsored by the White House Council on Environmental Quality, recently challenged data center operators to improve energy efficiency by 20 percent. However, out of approximately 3 million US data centers, the initiative announced the support of just 19 data center partners.

The Better Buildings Challenge suggests well-known industry tactics for improving energy efficiency, including implementing hot and cold aisles, improving management of cool air flow to servers, and adhering to ASHRAE’s recommended temperature range. However, the common – but misguided – notion that data centers must choose between uptime and efficiency when optimizing and allocating resources is keeping data centers from fully capitalizing on these tactics and other energy efficiency technologies.

It’s not difficult to see why: data center operators and owners are largely measured against uptime – that is, the 24/7 availability of data center services in a constantly connected, data-driven world. Consumers are quick to complain when a website is down, but rarely inquire about the energy efficiency of the data center behind it.

In order to accommodate the demand of always-on availability and the power and cooling infrastructure necessary to provide it, data centers are typically oversized in terms of capacity so that it’s easier to meet potential demands for increased services further down the road. Unfortunately, while this strategy drastically reduces overall energy efficiency, those responsible for data center uptime are often not in charge of actually paying the utility bill.

So how can data centers have both uptime and efficiency? By choosing the right efficiency measures and ensuring proper planning to reduce human error. These tactics will not only decrease energy consumption without forfeiting availability, but also reduce capital and operational expenses – an added bonus – while reducing risk. Capital investments into efficiency measures can even produce a full return on investment within just a couple years, or even months.

Choosing the right efficiency measures

Many of the tactics suggested by the Better Buildings Challenge are spot-on in terms of enabling data centers to improve energy efficiency without sacrificing uptime:

• Implementing hot and cold aisles can be done quickly and cheaply, greatly increasing energy efficiency.
• Improving airflow management through separation and containment also increases energy efficiency and does not take up extensive resources or time.
• By raising data center temperature to adhere to ASHRAE guidelines, cooling units are more efficient but infrastructure performance is unaffected.

However, there are other methods to consider as well, depending on the needs of the organization:

• Data center owners and operators can’t improve what they can’t measure; a data center infrastructure management (DCIM) solution can provide a baseline from which to measure energy efficiency while also increasing visibility into infrastructure.
• Zombie servers should be put to rest once and for all; they’re a drain on energy and waste valuable whitespace. Organizations can decommission underutilized servers easily by engaging with the right departments in identifying and removing them. For example, Barclays recently removed over 9,000 servers resulting in a $4.5 million reduction in the company’s power bill.
• Economizers, which take advantage of ambient temperature to cool a data center, can be a good investment for the right data center.
• Perfect rightsizing is impossible, but choosing components and designs with good partial-load efficiency will minimize the impact. During the design phase, consider the partial-load specs of uninterruptible power supplies (UPS), computer room air conditioners (CRAC) and computer room air handler (CRAH) units, chillers and pumps.
• Modular, scalable power and cooling equipment is ideal for both new data center builds and existing facilities because owners and operators can add capacity as needed, instead of having to overbuild in order to accommodate potential future demand (that may never come to fruition). Modular infrastructure is often standardized and prefabricated, making the acquisition, implementation and commissioning process painless and straightforward while also minimizing future risks to downtime and simplifying maintenance.
• Data centers with older uninterruptible power supply (UPS) systems should consider modernizing and rightsizing their infrastructure, which can drastically improve efficiency and ensure uptime during a power outage.

Proper Planning to Reduce Human Error during Implementation of Energy Efficiency Measures

Owners and operators can avoid most instances of downtime during the implementation of an efficiency measure by planning ahead and reducing the chances for human error. The following tips can be taken into account to ensure a smooth, error-free project.

• Any work being done that involves the electrical system is, of course, critical. Owners and operators should have maintenance and inspections performed on a regular basis. Particularly for data centers that do not have the kind of built-in redundancy that many larger enterprise facilities have, this is imperative.
• A somewhat simple – but effective – method for safeguarding existing systems during construction is to install a temporary wall blocking the worksite from the rest of the facility. Colocation providers often use this method when expanding their data centers.
• Maintaining proper communication and providing advance notice of data center upgrades is an easy way to avoid human error and make sure all parties involved are on the same page. Site visits, contingency plans and pre-meetings are all important elements to maintaining a well-orchestrated project.
• Only allow necessary personnel within the data center during construction or installation to reduce the chance of human intervention and error.

There is a strong business case for implementing energy efficiency measures, including savings in both operational and capital expenses. By only investing in the infrastructure they need to fit current capacity needs and reduce utility bills by implementing the methods provided data center owners and operators can realize numerous cost, time and energy savings – without sacrificing the uptime that’s so important in today’s 24/7/365 world.

Robert Bunger is a global solution architect for Data Center Solutions, Schneider Electric.