Better data centers through machine learning

It’s no secret that we’re obsessed with saving energy. For over a decade we’ve been designing and building data centers that use half the energy of a typical data center, and we’re always looking for ways to reduce our energy use even further. In our pursuit of extreme efficiency, we’ve hit upon a new tool: machine learning. Today we’re releasing a white paper (PDF) on how we’re using neural networks to optimize data center operations and drive our energy use to new lows.

It all started as a 20 percent project, a Google tradition of carving out time for work that falls outside of one’s official job description. Jim Gao, an engineer on our data center team, is well-acquainted with the operational data we gather daily in the course of running our data centers. We calculate PUE, a measure of energy efficiency, every 30 seconds, and we’re constantly tracking things like total IT load (the amount of energy our servers and networking equipment are using at any time), outside air temperature (which affects how our cooling towers work) and the levels at which we set our mechanical and cooling equipment. Being a smart guy—our affectionate nickname for him is “Boy Genius”—Jim realized that we could be doing more with this data. He studied up on machine learning and started building models to predict—and improve—data center performance.

The mechanical plant at our facility in The Dalles, Ore. The data center team is constantly tracking the performance of the heat exchangers and other mechanical equipment pictured here.

What Jim designed works a lot like other examples of machine learning, like speech recognition: a computer analyzes large amounts of data to recognize patterns and “learn” from them. In a dynamic environment like a data center, it can be difficult for humans to see how all of the variables—IT load, outside air temperature, etc.—interact with each other. One thing computers are good at is seeing the underlying story in the data, so Jim took the information we gather in the course of our daily operations and ran it through a model to help make sense of complex interactions that his team—being mere mortals—may not otherwise have noticed.

A simplified version of what the models do: take a bunch of data, find the hidden interactions, then provide recommendations that optimize for energy efficiency.

After some trial and error, Jim’s models are now 99.6 percent accurate in predicting PUE. This means he can use the models to come up with new ways to squeeze more efficiency out of our operations. For example, a couple months ago we had to take some servers offline for a few days—which would normally make that data center less energy efficient. But we were able to use Jim’s models to change our cooling setup temporarily—reducing the impact of the change on our PUE for that time period. Small tweaks like this, on an ongoing basis, add up to significant savings in both energy and money.

The models can predict PUE with 99.6 percent accuracy.

By pushing the boundaries of data center operations, Jim and his team have opened up a new world of opportunities to improve data center performance and reduce energy consumption. He lays out his approach in the white paper, so other data center operators that dabble in machine learning (or who have a resident genius around who wants to figure it out) can give it a try as well.

Posted by Joe Kava, VP, Data Centers

Going solar with SunPower

Just because Earth Day is over doesn’t mean we’re done doing good things for the planet. Yesterday we announced our biggest renewable energy purchase yet: an agreement with our Iowa utility partners to supply our data center facilities there with up to 407 megawatts of wind energy.

Today, we’re taking another step towards a clean energy future with a major new investment. Together with SunPower Corporation we’re creating a new $250 million fund to help finance the purchase of residential rooftop solar systems—making it easier for thousands of households across the U.S. to go solar. Essentially, this is how it works: Using the fund ($100 million from Google and $150 million from SunPower), we buy the solar panel systems. Then we lease them to homeowners at a cost that’s typically lower than their normal electricity bill. So by participating in this program, you don’t just help the environment—you can also save money.

A home sporting SunPower solar panels

SunPower delivers solar to residential, utility and commercial customers and also manufacturers its own solar cells and panels.They’re known for having high-quality, high reliability panels which can generate up to 50 percent more power per unit area, with guaranteed performance and lower degradation over time. That means that you can install fewer solar panels to get the same amount of energy. And SunPower both makes the panels and manages the installation, so the process is seamless.

This is our 16th renewable energy investment and our third residential rooftop solar investment (the others being with Solar City and Clean Power Finance). Overall we’ve invested more than $1 billion in 16 renewable energy projects around the world, and we’re always on the hunt for new opportunities to make more renewable energy available to more people—Earth Day and every day.

Posted by Rob Parker, Renewable energy team

Solar in California and Arizona: More of a good thing

You’d think the thrill might wear off this whole renewable energy investing thing after a while. Nope—we’re still as into it as ever, which is why we’re so pleased to announce our 14th investment: We’re partnering with global investment firm KKR to invest in six utility-scale solar facilities in California and Arizona. Developed by leading solar developer Recurrent Energy, the projects have a combined capacity of 106MW and will generate enough electricity to power over 17,000 U.S. homes. Google will make an approximately $80 million investment into these facilities.

The 17.5 MWac/22 MWp Victor Phelan project (pictured), located in San Bernardino, Calif., is part of six Recurrent Energy developed projects acquired by Google and KKR. The six-project portfolio is expected to operational by early 2014 and will generate enough clean electricity to power more than 17,000 U.S. homes.

This investment is similar to one we made back in 2011, when we teamed up with KKR and invested $94 million in four solar facilities developed by Recurrent. Those facilities have since started generating electricity, and we’ve committed hundreds of millions more—more than $1 billion in total—to renewable energy projects around the world.

These investments are all part of our drive toward a clean energy future—where renewable energy is abundant, accessible and affordable. By continuing to invest in renewable energy projects, purchasing clean energy for our operations and working with our utility partners to create new options for ourselves and for other companies interest in buying renewable energy, we’re working hard to make that future a reality.

Posted by Kojo Ako-Asare, Head of Corporate Finance

Another windy day in Texas: a new power purchase agreement

As part of our quest to power our operations with 100% renewable energy, we’ve agreed to purchase the entire output of the 240 MW Happy Hereford wind farm outside of Amarillo, Texas. This agreement represents our fifth long-term agreement and our largest commitment yet; we’ve now contracted for more than 570 MW of wind energy, which is enough energy to power approximately 170,000 U.S. households.

The Happy Hereford wind farm, which is expected to start producing energy in late 2014, is being developed by Chermac Energy, a small, Native American-owned company based in Oklahoma. The wind farm will provide energy to the Southwest Power Pool (SPP), the regional grid that serves our Mayes County, Okla. data center.

Some (happy) cows on the future site of the wind farm. 

 The cows will still have plenty of room to graze between the turbines.

The structure of this agreement is similar to our earlier commitments in Iowa and Oklahoma. Due to the current structure of the market, we can’t consume the renewable energy produced by the wind farm directly, but the impact on our overall carbon footprint and the amount of renewable energy on the grid is the same as if we could consume it. After purchasing the renewable energy, we’ll retire the renewable energy credits (RECs) and sell the energy itself to the wholesale market. We’ll apply any additional RECs produced under this agreement to reduce our carbon footprint elsewhere.

This type of power purchase agreement represents one of several ways we’re working to make additional renewable energy available for both our data centers and the communities in which we operate. In Scandinavia, due to the region’s unified power market and grid system, we’re able to purchase wind energy in Sweden and directly consume it at our Hamina, Finland data center. We’re also working with our local utility partners to develop new options. In 2012, we signed an agreement with GRDA, our utility partner in Oklahoma, to green the energy supply to our Oklahoma data center with 48 MW of wind energy from the Canadian Hills Wind Project. Earlier this year, we began working with Duke Energy to develop a new renewable energy tariff (PDF) in North Carolina.

We take a comprehensive approach to acquiring renewable energy for our operations. We’ll continue working directly with utility providers, collaborating with industry regulators and pursuing creative agreements (PDF) like the Happy Hereford PPA.

Posted by Matt Pfile, Senior Manager, Data Center Energy and Location Strategy

How green is the Internet?

More than ever, people are using the Internet to shop, read, listen to music and learn. And businesses rely on Internet-based tools to operate and deliver their services efficiently. The Internet has created all kinds of new opportunities for society and the economy—but what does it mean for the environment?

We’ve been working to answer that question and enlisted the help of Lawrence Berkeley National Laboratory (Berkeley Lab) to gather more data. Their study (PDF), released today, shows that migrating all U.S. office workers to the cloud could save up to 87 percent of IT energy use—about 23 billion kilowatt-hours of electricity annually, or enough to power the city of Los Angeles for a year. The savings are associated with shifting people in the workforce to Internet-based applications like email, word processing and customer relationship software.

These results indicate that the Internet offers huge potential for energy savings. We’re especially excited that Berkeley Lab has made its model publicly available so other researchers and experts can plug in their own assumptions and help refine and improve the results.

Of course, understanding the impact of shifting office applications to the cloud is only part of the story, which is why last week we hosted a summit called “How Green is the Internet?” to explore these questions in greater detail. At the summit, experts presented data on how the growth of Internet infrastructure, including devices like phones and tablets, can impact the environment. We also saw great excitement about the potential for entirely new Internet-enabled tools in areas like transportation, e-commerce and digital content to deliver huge energy and carbon savings. We’ve posted the videos from those sessions and invite you to take a look.



One of our goals in hosting the summit and supporting the Berkeley Lab study was to identify and encourage new research on this topic. We’ll continue to work to answer some of these questions, and we hope others will too.

Posted by Michael Terrell, Senior Policy Counsel, Energy & Sustainability

Powering our Finnish data center with Swedish wind energy

What do a Swedish wind farm developer, a German insurance company and Google’s Finnish data center have in common? As of today, a lot. We’ve just inked agreements with O2 and Allianz to supply our Finnish data center with renewable energy for the next 10 years—our fourth long-term agreement to power our data centers with renewable energy worldwide, and our first in Europe.

Here’s how it works: O2, the wind farm developer, has obtained planning approval to build a new 72MW wind farm at Maevaara, in Övertorneå and Pajala municipality in northern Sweden, using highly efficient 3MW wind turbines. We’ve committed to buying the entire output of that wind farm for 10 years so that we can power our Finnish data center with renewable energy. That agreement has helped O2 to secure 100% financing for the construction of the wind farm from the investment arm of German insurance company Allianz, which will assume ownership when the wind farm becomes operational in early 2015.

This arrangement is possible thanks to Scandinavia’s integrated electricity market and grid system, Nord Pool. It enables us to buy the wind farm’s output in Sweden with Guarantee of Origin certification and consume an equivalent amount of power at our data center in Finland. We then “retire” the Guarantee of Origin certificates to show that we’ve actually used the energy.

As a carbon neutral company, our goal is to use as much renewable energy as possible—and by doing so, stimulate further production. The Maevaara wind farm not only allows us to make our already highly energy-efficient Finnish data center even more sustainable, it also meets our goal of adding new renewable energy generation capacity to the grid.

Of course, using renewable energy is good for the environment, but it also makes long term financial sense. That’s why, in addition to protecting ourselves against future increases in power prices through long-term purchasing for our operations, we also invest in new renewable energy projects that will deliver a return for our money. In recent years we’ve committed more than $1 billion to such projects in the U.S., Germany and, just last week, South Africa. We’ll continue to look for similar opportunities around the globe.

Posted by Francois Sterin, Senior Manager, Global Infrastructure Team