For love of money and the environment

A discussion about dirty data centers

Facebook's planned data center in Fort Worth, Texas

Facebook's planned data center in Fort Worth, Texas.

Credit: Facebook

The Paris climate talks just concluded, with at least one mainstream media source hailing the resulting agreement as “the world’s greatest diplomatic success.” Back home, aside from those married to the issue, we hardly noticed, especially those of us in the technology industry. Life continued as normal, we drove to work and back home in our SUVs, spent the evening engaged in some form of electronic entertainment and then went to bed in our climate-controlled homes, while miles away our companies' data centers hummed through the night, powering the Web for millions of night owls.

Why didn’t this news create a bigger ripple? Why does it appear in The New York Times but not in many technology publications? Why isn’t an influential tech journalist like David Pogue or Farhad Manjoo putting pen to paper (or fingers to keys?) and analyzing the impact of this agreement on the technology industry? Is the technology industry immune to this conversation, or are its players simply too busy keeping the lights on to take on projects that don’t impact today’s bottom line?

According to some data published last year, large data centers can often use as much electricity as a small U.S. town. That statistic in itself should remove any notion of immunity. It is high time that we, as technology leaders, began to strive for more sustainable solutions and practices. Our waste of energy is not only bad for the planet; it can cripple the balance sheet as well.

A simple place to start is with data centers, which are responsible for 17% of the entire carbon footprint created by the technology industry. Some large corporations, including Facebook and Google, have pledged considerable sums of money toward establishing renewable energy sources for their massive data centers. But not every company has the wherewithal to do that. The majority of CIOs and CTOs I talk with are wondering what they can do that is right for their environment. They don’t have a billion dollars to pledge toward renewable energy and so they need to be more creative.

So what’s a CIO or CTO to do? I sat down with Julian Kudritzki, COO of Uptime Institute, a "data center industry advisory organization." We discussed cost effective steps a business can take to reduce the carbon footprint and improve the efficiency of its data centers. Here’s the result of that conversation:

What are some easy things a CTO, COO or sustainability executive can do today to make his/her data center(s) more energy-efficient?

Here are five ideas for those stakeholders to consider:

1. Involve yourselves as leadership. Any IT efficiency initiative is going to be short-lived and less effective without executive authority to drive the changes across the organization. For both one-time and sustained savings, executive leadership must address the management challenges inherent in any process improvement. Those organizations that have been successful in this regard have shared the cost and resource consumption.

2. Implementing a DCIM (data center infrastructure management) could impact efficiency and costs. DCIM is a software suite that collects and manages information about a data center’s assets, resource use and operational status. Many IT organizations have purchased or are in the procurement process with a DCIM software implementation. Yet only 17% of enterprise organizations expect a return on that investment within two years, and half do not expect to achieve ROI on a DCIM purchase at all. What is driving this dysfunction? DCIM buyers are misusing these sophisticated, expensive tools to replace existing facilities management functions. Over 75% of the usage scenarios for DCIM revolve around space planning and environmental monitoring — ostensibly functions that most data center facilities already had in place. And yet less than 25% of DCIM buyers are using the tool to improve server utilization rates, implement chargeback or to conduct financial analysis. The cost and time required to fully implement DCIM can be a barrier, but it is well worth considering — especially for large organizations.

3. Server decommissioning. Decommissioning old or inefficient equipment has several direct cost-saving benefits, in addition to reducing risk and complexity. In fact, leading IT organizations such as Barclays, Visa and AOL have saved tens of millions of dollars implementing disciplined server decommissioning programs. However, in our most recent annual survey of senior IT infrastructure stakeholders, 45% have no scheduled auditing to identify and remove unused machines, so there are potentially huge savings going unnoticed. How do you know what much you are wasting if you don’t measure it? 

4. Quit focusing on power usage effectiveness (PUE), a ratio that represents how much of a data center's energy use is consumed by IT equipment versus the supporting facilities' infrastructure. PUE is useful for tracking the efficiency of a data center's cooling systems over time and seasonal variables, but it's a lousy management tool, as PUE only addresses the least impactful aspect of IT energy usage. Yet, our research shows that 82% of IT executives track PUE and report that metric to their corporate management. By focusing on PUE, IT executives are spending effort and capital for diminishing returns and ignoring the underlying drivers of poor IT efficiency. PUE tells you nothing about asset utilization, and a PUE could look great for a data center that is packed with comatose servers working hard to squander expense and deliver no business value.


5. Adopt IT chargeback. In simple terms this means that any departments using IT services (including data centers) are charged directly for the cost of those services. Chargeback tends to reduce overall resource consumption as business units stop hoarding surplus servers or other resources to avoid maintaining underutilized assets. Perhaps most importantly, IT chargeback drives cost control as users become aware of the direct costs of their activities. This can extend the life of existing resources and infrastructure, make it possible to defer resource upgrades, and identify underutilized resources that can be deployed more efficiently.

What exactly is a comatose server, and how is it affecting my data center’s energy efficiency?

A server is classified as being comatose if it is not delivering information or computing services to users. Within large organizations, we typically see that at least 20% of servers are comatose: they are idle, obsolete or unused, but are still plugged in, “on,” and drawing power. Wasteful, or comatose, servers hide in plain sight in even the most sophisticated IT organizations. These servers represent a triple threat in terms of waste — they squander power at the plug, waste data center facility capacity and space, and incur software licensing fees and hardware maintenance costs.

What percentage of my budget should I set aside in 2016/2017 to clean up my datacenters?

IT efficiency is managed, not bought. We have found that efficient IT is a function of organizational coherence rather than expense outlay. It really comes down to identifying areas for improvement and getting the executive buy-in to make the changes necessary to achieve those goals. Get your organization's attention by using the average 20% comatose server figure as a benchmark, there’s clearly a huge potential upside to devoting resources to weeding out inefficiency. The process of analyzing ownership, utilization and cost of comatose servers alone will identify opportunities for organizational improvement, as well as save significant expense. Then, you will be asked where else you can save in IT services delivery. Use your clout to improve purchasing behaviors, new technology evaluation and adoption, interdisciplinary initiatives, etc.


There you have it. The time is now to take some of 2016’s budget and apply it to sustainability from an energy and cost standpoint. We all have the ability to create the most innovative and meaningful technology for our customers, but if we continue to waste energy there will be no customers to sell to, no power for our data centers and ultimately no need to fire up that supercomputer. The time to make this investment is now. Our leaders have spoken. 

This article is published as part of the IDG Contributor Network. Want to Join?

The march toward exascale computers
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies