Jonathon G. Koomey’s recently released and highly publicized report, “Growth in Data Center Electricity Use: 2005 to 2010” is nothing like other reports that have warned that power consumption in data centers is out of control. In the past five years, we have seen article after article extrapolating from an earlier Koomey publication and the U.S. Environmental Protection Agency’s (EPA) Report to Congress, suggesting that total U.S. data center energy consumption is now at 3 to 5 percent or more of total U.S. use.

In his latest update, Koomey, a Ph.D. and consulting professor at Stamford University, found that the electricity used in global data centers accounted for between 1.1 and 1.5 percent of total electricity use. In the U.S., the numbers rise to between 1.7 and 2.2 percent. This is far below the expectations set by the EPA.    

Of the five infamous growth curves plotted by the EPA, the middle plot would seem to have best described the growth in energy use. The EPA labeled this curve the “Improved Operation Scenario.” However, numerous reasons other than improved operations account for the slowed growth. Obviously, the economy has had an impact, but technologies have changed as well. The growth of volume servers and mid-range processors has slowed also, but high-end servers, where we typically get more processing per watt, still grew. So surely we can now expect the virtualization and cloud promoters to start claiming credit for impacting power growth in a positive way.

Indeed they have, but there were numerous other reasons for the limited growth, including the focus on ridding data centers of ghost servers, use of power management applications in servers, and the rapid increase in storage densities.

Given the five-year focus on efficiency plus a lagging economy, the moderate growth did not surprise me. I was surprised by Koomey’s reference to the energy used by Google’s data centers.  What surprised me was not the power consumed by Google (less than 1 percent of all data center energy), but that Koomey thought it necessary to reference one user.

This single reference struck me as more the old politic game of pick on the big guy so all the little guys can slip by. So I went looking for other similar references on the industry leaders and sure enough, they weren’t hard to find. 

I found a summary of a report titled, “Carbon Strategy Benchmark: Internet Sector,” through a LinkedIn discussion group. The report, available via subscription at Verdantix, an independent UK-based analyst firm, is a review of the carbon footprints of 14 of the largest global internet companies. The report summary found that Akamai, Applem, and eBay outpaced their competitors on energy efficiency but found Amazon, Expedia, and Tencent to be carbon-reporting laggards. Further, it went on to say that none of the 14 global internet giants covered had their data validated by an independent greenhouse gas (GHG) auditing firm.

What I find wrong here is the focus on the 14 largest firms, which combine for only a fraction of the total global electricity used by data centers. These are all major firms with a strong economic incentive to minimize their electricity/carbon footprints, and they work on that every day. What about the other 95+ percent of the industry, which makes little effort to control their electrical/carbon footprints?

In the last several years I have been in data centers where efficient water-cooled mechanical systems were replaced by straight direction expansion (DX) systems because the owner did not want to manage water-based cooling towers. Another site included a data center in a high-rise with a glass curtain wall facing due east to catch the morning sun. In winter, lacking any vapor barrier, its infrared humidifiers ran full out to maintain humidity levels.

I have also met a colo operator who used an uninsulated concrete block building to house his operation to keep building costs down. When I asked him about the energy wasted because of the building shell, he simply stated that power was a pass-through to his clients as an operating expense, so it didn’t affect him. Surprisingly, his 10,000 square foot site was full, and he was looking to expand. 

While it is always good to rate the top firms, I have a hard time seeing them as laggards because they tend to be far better than the rest of the industry. What we really need is a more global focus on best practices and better benchmarking of consumption and carbon footprinting. The European Union has gone a long way in prescribing performance goals, but the rest of the world needs to catch up. This catch-up does not require new laws to be passed; it just needs leadership and a sense of social responsibility to make it happen. In the long run laws just hold us back in the data center industry because they get written around current technologies. As technologies advance the old laws seldom keep up.

Almost all the new data centers built in the last five years track their energy consumption, but what is done with all this valuable data? It goes nowhere. What we need is an organization that can collect everyone’s data on a daily basis, aggregate it, and reported back as to how the data center performed each month as compared to its peers. If we sincerely care about efficiency, we cannot wait for Jonathan Koomey to write another report in 2015. We need good accurate data, and we need it now. Only by banding together will we make the industry the best it can be. 


Reprints of this articleare available by contacting Jill DeVries at or at  248-244-1726.