Machine Learning endorsing "Green Computing" by optimizing energy efficiency of data centers.

Compare to traditional computing, cloud computing store massive data centrally ,remotely by allowing multi-tenancy using visualization technologies. This lowers carbon emission due to high energy efficiency.

But still cloud computing worldwide consumes 2 percent of global energy demand, and this is more than the power consumed by India or Germany. This consumption is growing at a rate of about 12 % per annum. Another challenge is more than half of this energy consumed from non-renewable sources like coal & nuclear power.

So, the bigger question is "What can be done to make world data centers more "Green Data Centers". What Can be done to increase power usage effectiveness across all data centers.

Google developed a set of machine learning algorithms to create data model on top of data center operational equipment and energy /cooling devices' sensor data to model DC performance and improve energy efficiency. Morden large scale data centers have its operational equipments (mechanical and electrical ) and their sensors throw every day millions of data points. And this data points analysed by machine learning algorithms also known as "neural network framework", to predict power utilization efficiency. This framework had been tested at Google's own data center, and energy efficiency improvement has been proven as part of the result .

Below link provides more detail about Google's this Machine learning model for data centers energy optimization

http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42542.pdf