Cloud Computing Creates Opportunities for Greater Efficiency!

Brian Janous, Utility ArchitectBrian Janous, Utility Architect
Data Center Advanced Development

The New York Times has launched a series looking at the environmental and economic impacts of cloud computing. These are critical issues to all cloud users and providers, and areas in which Microsoft has dedicated substantial resources, helped create industry metrics for, and publicly shared our best practices on for many years. We may not agree with all of the conclusions that the Times reaches, but we appreciate their desire to focus attention on these important topics.  

I want to share my thoughts on a couple areas The Times reported on. I joined Microsoft's Data Center Advanced Development team in 2011 to continue to accelerate the company's focus on data center resource requirements including power consumption, water usage, and carbon emissions, so these issues are core to my work.

Before jumping in, I want to start with our most fundamental beliefs about cloud computing and data centers. At Microsoft, we believe that cloud computing has the potential to revolutionize how people everywhere work, play and communicate. We also believe that cloud computing, combined with continued innovation in data center design and operations, can reduce the energy consumption and environmental impacts of the technology the world relies upon for productivity, communication and entertainment. When it comes to energy consumption, environmental sustainability, and strong communities, we believe online services are not the problem; they are actually a huge part of the solution.  And we are committed to working with the industry, customers, policymakers, and local communities to ensure that we achieve the full promise of cloud computing in a safe and sustainable way.

The first installment of the Times series looked at the energy usage of a wide range of data centers around the world. While the article identifies the key issues of efficiency and utilization, it conflates these two key but distinct challenges, wrongly implying that underutilization of servers in data centers equates to unneeded consumption of massive amounts of energy.  In doing so the article fails to recognize that not all data centers are created equal, nor are the operations and software applications running inside those data centers equally utilized. Several other news and industry observers have raised questions about some of the analysis, in particular the way the article doesn't draw distinctions between modern, state-of-the-art data centers managed by global cloud service providers and smaller, older data centers managed by corporate IT departments.

By way of background, efficiency is a measure of how well the system operates when it is working. If a server isn't working, even the oldest data centers have technology to adjust their consumption of electricity and water. In contrast, utilization measures how often the system is performing useful work. For example, if I am considering the purchase of a car and efficiency is my chief concern, I may opt for a hybrid. However, if I determine that I will only be driving or utilizing the car a few times a month, a car-sharing service might be a better fit.

At Microsoft we are focused on both issues, and are working to provide efficiencies that were unheard of a few short years ago. We are on a journey where the innovation rate is staggering. Indeed, the fundamental promise of the cloud and the reasons that Microsoft has invested so heavily in evolving the data center infrastructure and the software applications running in them is that we believe our cloud services can help improve both efficiency and utilization. To continue the car analogy, you could say that with our cloud services we're offering enterprise customers a hybrid car-sharing service that gives them access to the most efficient computing power available, and lets them use only the resources they need, when they need them.

The constraints experienced by the thousands of corporate IT departments, that result in inefficiency and underutilization, cannot be resolved without a paradigm shift similar to the evolution of sending a letter via a mail truck that took days to arrive (that is costly, time consuming and uses a lot of carbon) to clicking the send button and instantly delivering thousands of letters to customers today. The efficiency of our modern, right-sized, Generation 4 data centers enables the high utilization possible in a globally distributed cloud environment.  

A few years ago Microsoft commissioned a study by Accenture and the environmental consulting firm WSP that looked at the total energy savings resulting when organizations move common business applications such as Microsoft Exchange for e-mail from their own servers to servers hosted by Microsoft and other cloud operators. The study found that large enterprises can expect to cut their energy use per user by at least 30 percent and small business up to 90 percent when moving to the highly optimized and efficient cloud data centers we operate. Microsoft's latest data centers use about 50 percent less energy than those from three years ago, and only one to three percent of the water used by traditional data centers in the industry.

Additionally, by migrating applications from physical to virtual machines and consolidating these applications onto shared physical hardware, Microsoft data centers are increasing utilization of server resources such as central processing unit (CPU), memory, and disk input/output. Microsoft has been using technologies such as Hyper-V to increase virtualization and thus utilization year over year, which in turn helps increase the productivity per watt of our operations.  Microsoft is also actively working on broad-based adoption of Microsoft's Windows Azure cloud operating system, which uses virtualization in its core.  On Windows Azure, an application typically has multiple instances, each running a copy of all or part of the application's code. Each of these instances runs in its own virtual machines, with a hypervisor specifically designed for use in the cloud.

Microsoft has also been on a path of continual improvement with respect to the energy consumption in our data centers. Energy is a core concern, and one of the largest operating expenses for our data centers. Although we have achieved substantial reductions in the energy required to provide reliable, cost-effective on-line services, we continue to invest millions in driving further innovation in this area. While inefficiencies and waste may get lost in enterprise IT environments, I can assure you they are not lost at Microsoft. Our customers expect reliable, secure cloud and online services and we are committed to providing those services with sustainability, efficiency and cost in mind.

The second Times installment looked at the town of Quincy, Washington, home to a number of large data centers, including those operated by Microsoft, Yahoo, Dell and others. The article uses what we believe are a few isolated situations to paint a negative picture of the relationship between big data center operators, including Microsoft, and the local community.

We are proud of our relationship with Quincy.  Microsoft has created over 50 good-paying permanent jobs in Quincy, and hundreds more construction jobs.  Our property taxes have helped fund community services, including an expansion of the local library and a new fire station. We built a multi-million dollar water treatment plant at Quincy, and leased it to the city for 30 years for $10 per year last year. We operate our facilities safely and we follow all applicable laws and regulations, but even more importantly we are part of the Quincy community.

One section of the article implies that Microsoft has run its diesel backup generators in excess of what is required to provide safe, reliable power to our data centers. We would respectfully disagree. Diesel generators are a costly alternative to grid-supplied power. The cost to run generators is several times higher than power purchased from the grid. And that doesn't account for the costs associated with purchasing, installing, and maintaining the generators. As we have previously stated, we are working to create ever-greater efficiency, power supply reliability, and system resiliency in our data centers and software applications, with the goal of significantly reducing or even eliminating the use of generators entirely. We are already doing this in our Chicago data center, where our container bays are not backed up with generators. In addition, the newest phase of our Boydton, Virginia facility is designed without any generators, a reflection of more fully resilient applications running there. We will continue trending in this direction. One of the main reasons we can do this is because of the scale at which our operations run.

These are important issues, and we will continue to work with industry, governments, customers, local communities, and other stakeholders to improve the efficiency, sustainability, reliability, and safety of our data centers, and to share our best practices with others.  It's great to dream about what lies just beyond the horizon and work towards helping our customers achieve greater efficiencies as well. The pace of technology change makes the ability to analyze our industry even tougher today than it used to be. But our ability to deliver everything I discussed above is a reality today. Our web-scale cloud infrastructures and software applications has fueled an incredible transformation in the workplace, delivering ever-more powerful tools that have increased productivity, streamlined communications, and improved our ability to help over a billion customers and 20 million businesses in over 76 markets worldwide find and use information. Our fundamental belief that technology can change the world and improve people's lives is why we do what we do.

You can read more about our best practices for cloud data centers in published videos, papers, and blogs available on our web site, please visit www.globalfoundationservices.com.

 

Brian Janous, Utility Architect, Data Center Advanced Development