Google's data centers are greener than ever

Google has revealed that its data centers are twice as energy efficient as a typical enterprise data center thanks to more than a decade worth of work spent on improving their energy usage and cooling.

In addition to being the world's largest corporate purchaser of renewable energy, the search giant has also taken action on climate change by minimizing the amount of energy it uses in the first place.

In a blog post detailing Google's work on making its data centers more energy efficient, the head of engineering at Google Cloud, Urs Hölzle pointed to a new paper in Science which validates the company's efforts. The study found that efficiency improvements have kept energy usage almost flat across the world's data centers, even as demand for cloud computing is at an all time high.

The paper, titled “Recalibrating global data center energy-use estimates”, shows that while the amount of computing done in data centers increased by close to 550 percent between 2010 and 2018, the amount of energy consumed by data centers only grew by six percent during the same time period. While data centers now power more applications for more users than ever before, they still account for about one percent of global electricity consumption, which is the same proportion as in 2010.

Improving efficiency

Over the past few few years, Google has been working relentlessly to eliminate waste at every level of its operations in order to improve efficiency at its data centers. The company designed its highly efficient Tensor Processing Units, which are the AI chips that power its advances in machine learning, and outfitted all of its data centers with high-performance servers. 

Then beginning in 2014, Google began employing machine learning to automatically optimize cooling in its data centers. At the same time, the company deployed smart temperature, lighting and cooling controls to further reduce the energy used at its data centers.

These efforts have yielded promising results and compared with five years ago, the company now delivers around seven times as much computing power with the same amount of electrical power. Google's AI-powered recommendation system, which directly controls data center cooling, is also delivering consistent energy savings of around 30 percent on average.

Hyperscale data centers are far more energy efficient than local servers and because of this, a person or company can immediately reduce the energy consumption associated with their computing by simply switching to cloud-based software.




via Tech Trade

Comments