Updated: Jun 28, 2018
We must think about progress in deploying new infrastructure
It almost seems like piling on to attack the NY Times assessment of datacenters and their impact on society at this point. When Forbes, InformationWeek, GigaOM, Wired and tons of other publications point out the weakness in the Times article, why add fuel to the fire? Because they are missing an enormous piece.
You see data taken as a point in time always misses a vitally important context, movement. If you take a point in time of the National Debt during the Clinton Administration, the number in abstract seemed shocking, but the trend was positive back then, eliminating deficits and targeting debt reduction. If you looked at crime statistics in NYC near the start of the Giuliani era, the raw numbers looked bad, but the trend was positive after implementation of the “broken windows” policy. Similarly, looking at datacenter energy consumption without context (not even going into the 2006-era data that was used), is not only a bad idea, but it paints the industry in such a light that true progress, continued progress, is threatened by the very piece that was penned presumably to move this industry forward.
I will not point out the logical fallacies, old data, cherry picked information to provide the allure of depth issues with the Times piece, but I will point to a vitally important piece that is not covered as deeply as needed and provide a bit of color from my career which can provide some insight into the progress the data center industry has made and how ill-timed and wrong-minded the Times piece is.
The Industry Cares about PUE (and you)
I was actually a bit shocked that the treatment of energy use in data centers seemed to completely avoid the industry’s use of standard measures to help improve utilization. The industry itself, without help from the NY Times or local, regional or national governments got together to develop metrics that would allow them to better control data center power consumption. They developed the PUE (Power Usage Effectiveness) and DCiE (Data Center infrastructure Efficiency) its reciprocal. I want to stress that reliance on these measures is not sufficient to cover advances in energy utilization. As with all metrics, this should be utilized as a way to engage in a conversation of efficiency not a simple target to be accomplished.
That being said, the industry’s embrace of PUE is an indicator of something very powerful. And that is competitive advantage. The data center industry has embraced the outcomes of decreased PUE, and more important holistic energy use control, as a competitive differentiator inside a highly competitive marketplace. There is nothing more powerful in society than a highly competitive marketplace, targeting a measurable output, to put powerful pressure on that metric moving forward. It is one thing for an industry to try to avoid bad press, negative consumer sentiment or regulation. But for those of us who have worked in these environments, there is nothing like a smart, agile competitor working to put you out of business to motivate your company to provide better energy utilization across the board. Bottom, and top, line control is vitally tied to decreased PUE in this industry as other competitive advantages are sometimes more difficult to land in a customer. Therefore, over the past few years, this industry all by itself has been motivated not simply out of benevolence but out of self preservation and competition to address the problems discussed in the Times articles.
My friends in the Central Governments and Local and Regional Governments of the world often talk about their efforts as anchor tenants in the future. They have scale unheard of and they have the possibility of pushing further an industry that has done incredible things in five short years to improved upon energy efficiency in data centers across the globe
And how have they done? Overwhelmingly well. Over at DataCenter Knowledge you can read about incredible advances in PUE ratings that were also coupled with build-out costs that were MORE efficient than previous efforts. I know that the scuttle can be about folks misrepresenting the metric and greenwashing things, but understand that this effort lies within an industry that is self motivated by competitive advantage to provide its customers with real, reliable and true data on energy use. And data that is easily proven once the data centers are commissioned and complete. The market will resolve to the most efficient use regardless of pressure from the press.
And unlike reporters, I speak from direct knowledge. As CIO for the State of Wisconsin I worked to authorize and build an enterprise class, Tier III data center for our state government. I was subject to PUE claims from all types of vendors. I saw it from our design/build firms, our PDU vendors, arguments from competing hardware vendors, and conversations with our utilities. We did not build this in a right-leaning anti-green location. We built it in Madison, Wisconsin. Arguably one of the greenest cities in North America. And competition allowed us to select for energy conservation and to position ourselves better than we were before.
The Largest Movement is to Come (Government)
Which leads me to my final point. Since working for Governor Doyle as his CIO, I have worked for a few places with an incredible impact on Datacenters, one of which was the center of the second Times article. I was lucky enough to have worked for Microsoft as their GM for Worldwide Government, and currently work as SVP in Salesforce.com’s Public Sector group. I have seen firsthand how the bifurcated data center market cares directly about energy usage and despite the challenges hurled at each in the Times article, I need to testify in defense of the customer side of data center practice.
I worked globally to push utilization of the public cloud at both companies. I was, and continue to be, impressed at how seriously each takes PUE and more advanced metrics in terms of energy use. Both companies press hard on their data center providers, both internal and external, to control energy utilization. First as an effort to control costs, but more importantly, as outlined above, as a competitive differentiator. I was part of multiple efforts to push for better PUE in order to leverage the data in an effort to more effectively move the market from internal on-premise data utilization to cloud based usage (including working directly with Microsoft’s first Chief Environmental Architect). A trend that allows for incredible scale and more pronounced positive effect on PUE than any other shift. As client companies decommission their internal data centers and push them to optimized and optimizing cloud based data centers, PUE is dramatically reduced across the entire data center universe. And it is pushed by competition, not external pressure. The more these efforts are embraced, and the benefits of multi-tenant cloud services are substituted, the more you will see all of us utilize data more efficiently. Especially when you consider the rising number of cloud based startups that are able to take advantage of their larger predecessors like Salesforce that have made cleaner, lower-PUE data centers available due to their anchoring effect on the market.
You see, the Times data was point data. Seemingly shocking. But did not take into account where IT load was TAKEN from. From inefficient “data centers” like the Facebook closet referenced in the first article. From unvirtualized environments stacking up 20% utilization rates. From siloed departments within one company being leveraged into an enterprise class data center to serve the global efforts of that same company. You see, the movement has been away from those inefficient structures into highly efficient, and competitive, structures that place incredible downward pressure on things like PUE, and even better holistic measures, moving forward.
And the huge risk of demonizing the data center community right now, is the big domino has yet to fall completely. The government space worldwide accounts for an unprecedented amount of compute capacity and data storage. I know, I managed it. If the efforts of this brave community are set out as demons, the pressure will resolve and advancement stalled at the worst possible time. As the governments of the world currently stand ready to make an unprecedented shift to cloud computing, now we must highlight the good work, reward those pushing harder and make the case that the governments of the world need to shift their compute methodology to take advantage of not just the incredible costs savings evident in cloud computing and consolidated data centers, but to also embrace a much more positive effect on the environment as we start to shift more fully into a data consumer based society.
My friends in the Central Governments and Local and Regional Governments of the world often talk about their efforts as anchor tenants in the future. They have scale unheard of and they have the possibility of pushing further an industry that has done incredible things in five short years to improved upon energy efficiency in data centers across the globe. I hope that the onslaught of press coverage unearthing the challenges to the Times article continue to embolden the data center industry to continue its pursuit of energy efficiency with the competitive advantage argument propelling them forward despite the efforts to introduce unneeded external pressure upon them.