How to Quench Data Centers' Thirst for Power (Op-Ed)
Credit: Igor Zh. | Shutterstock

Pierre Delforge, NRDC’s director of high tech energy efficiency, previously worked for 20 years in the IT industry in software development, hardware integration, and energy efficiency and climate programs. He contributed this article to Live Science's Expert Voices: Op-Ed & Insights

All online activity — from email and Internet use to social media and business — streams through nearly 3 million data centers across America, from small closets and large server rooms to mammoth "cloud" server farms. The centers' explosive growth is gulping huge amounts of energy, and despite some efficiency improvements, much of it is still wasted.

Although well-known Internet brands like Apple, Facebook, Google and others rightly pride themselves on the ultra-high efficiency of their immense data centers delivering search, social networking and other digital services to consumers and businesses alike, according to a new report from NRDC and Anthesis, these cloud-server farms are responsible for less than 5 percent of total data-center energy consumption in the nation, and are not representative of how the average U.S. data center operates.

Our study shows that many small, mid-size, corporate and multi-tenant data centers still waste much of the energy they use. Many of the roughly 12 million U.S. servers spend most of their time doing little or no work, but still drawing significant power — up to 30 percent of servers are "comatose" and no longer needed, while many others are grossly underutilized. However, opportunities abound to reduce energy waste in the data-center industry as a whole.  Technology that will improve efficiency exists, but systemic measures are needed to remove the barriers limiting its broad adoption across the industry. 

How much energy do data centers use?

The rapid growth of digital content, big data, e-commerce and Internet traffic more than offset energy-efficiency progress, making data centers one of the fastest-growing consumers of electricity in the U.S. economy, and a key driver in the construction of new power plants. If such data centers were a country, they would be the globe's 12th-largest consumer of electricity, ranking somewhere between Spain and Italy.

In 2013, U.S. data centers consumed an estimated 91 billion kilowatt-hours of electricity. That's the equivalent annual output of 34 large (500-megawatt) coal-fired power plants — enough electricity to power all the households in New York City, twice over, for a year. 

Meanwhile, our report projects data-center electricity consumption to increase to about 140 billion kilowatt-hours annually by 2020, requiring the equivalent annual output of 17 new power plants, costing American businesses $13 billion annually in electricity bills and emitting nearly 150 million metric tons of carbon pollution  annually.

Our last report, "Is Cloud Computing Always Greener?" found these smaller data centers have generally made much less progress than their hyper-scale cloud counterparts. Our latest analysis shows energy-efficiency advances are being hampered by persistent issues and market barriers, such as lack of metrics and transparency, and misalignment of incentives (i.e., the person who makes the decisions affecting efficiency is rarely the same person paying the energy bills).

Fixing the problem

While current technology can improve data center efficiency, we recommend systemic measures to create conditions for best-practices across the data center industry, including:

If you're a topical expert — researcher, business leader, author or innovator — and would like to contribute an op-ed piece, <a href=mailto:expertvoices@techmedianetwork.com>email us here</a>.
If you're a topical expert — researcher, business leader, author or innovator — and would like to contribute an op-ed piece, email us here.

 

  • Adoption of a simple, server-utilization metric. One of the biggest efficiency issues in data centers is underutilization of servers. Adoption of a simple metric, such as the average utilization of the server central processing units (CPUs), is a key step in resolving the energy-consumption issue. Measuring and reporting CPU utilization is a simple, affordable and adequate way of gauging data-center efficiency that could immediately drive greater IT energy savings. 
  • Rewarding the right behaviors. Data center operators, service providers and multi-tenant customers should review their internal organizational structures and external contractual arrangements and ensure that incentives are aligned to provide financial rewards for efficiency best practices. Multi-tenant data center stakeholders — those served by a single facility where they lease space, power, Internet connectivity, etc. — should develop a "green lease" contract template to make it easier for all customers to establish contracts that incentivize, rather than stand in the way of, energy savings.
  • Disclosure of data-center energy and carbon performance. Public disclosure is a powerful mechanism for demonstrating leadership and driving behavior change across an entire sector. In their corporate and social responsibility reports, industry leaders in data-center efficiency should voluntarily disclose operational performance metrics, such as fleetwide server utilization levels, and organizational performance (e.g., how they address split incentive issues internally and externally). 

If just half of the technical savings potential for data-center efficiency that we identify in our report is realized (taking into account market barriers), electricity consumption in U.S. data centers could be cut by as much as 40 percent. Today, that would represent a savings of 39 billion kilowatt-hours of electricity annually — equivalent to the annual electric consumption of nearly all the households in Michigan, improvements that would save U.S. businesses and their customers a whopping $3.8 billion a year.

Follow all of the Expert Voices issues and debates — and become part of the discussion — on Facebook, Twitter and Google +. The views expressed are those of the author and do not necessarily reflect the views of the publisher. This version of the article was originally published on Live Science.