Google Uses AI To Cool Data Centers and Slash Electricity Use
By Dan Heilman / CIO Today. Updated July 20, 2016.
Data centers are notorious for how much energy they require. Now, however, Google thinks it has found a way to mitigate that problem. Over the last few months, the tech giant has been using an artificial intelligence (AI) system built by DeepMind to control certain parts of its data centers to make its vast server farms more environmentally friendly.
DeepMind is the British AI company that Google bought for more than $600 million two years ago. The company is famous for developing AI technology that was able to beat the world’s best player of the board game Go. Although the company has worked on projects for various U.K. healthcare companies, it hasn’t yet turned a profit for Google.
40 Percent Cut in Energy Use
The use of the AI technology is "a phenomenal step forward" to help cut down energy usage in data centers, DeepMind research engineer Rich Evans and Google data center engineer Jim Gao said on Google's blog. In fact, the electricity consumption of data centers is on a pace to account for 12 percent of global electricity consumption by 2017, according to predictions in a 2015 report from Greenpeace. Some of the world’s biggest tech companies, including Apple, Amazon and Google, run some of the biggest data centers.
The DeepMind AI system has been used so far to curb energy consumption by Google’s data center cooling units, which stop the company’s servers from overheating. The coolers’ energy usage has been cut by up to 40 percent so far, which Google said has helped one of its data centers reach a 15 percent reduction in power usage efficiency (PUE).
PUE is the ratio of the total building energy used by equipment, such as pumps, chillers and cooling towers, to the energy used by IT equipment such as Google's servers. Google said it also plans to use DeepMind's AI across other parts of its data center infrastructure.
All About Optimizing
How did Google do it? The energy reduction was achieved by training DeepMind's self-learning algorithms to predict how hot data centers were going to get within the next hour, according to Evans and Gao. Armed with that data, the coolers were only able to run at the maximum temperature necessary to keep the servers sufficiently cool. Google’s data centers are used to run such services as Search, YouTube and Gmail.
Using a system of neural networks that zero in on different operating scenarios and parameters within the data centers allows DeepMind to make a more efficient and adaptive framework to understand data center dynamics and optimize efficiency, according to Evans and Gao.
"The implications are significant for Google’s data centers, given its potential to greatly improve energy efficiency and reduce emissions overall," Evans and Gao said. "This will also help other companies who run on Google’s cloud to improve their own energy efficiency."
Similar deployments of AI are becoming more and more common, Dave Schubmehl, research director for IDC’s Cognitive Systems and Content Analytics division, told us today.
"It’s all about optimization," said Schubmehl. "It can be used in supply logistics, shipping logistics and dynamic pricing in addition to keeping an industrial area at the right temperature. We’ll be seeing AI being applied to a lot more areas."