Slot 4

Data Center Power Management: Green, Efficiency, and Beyond
Shaolei Ren, University of California, Riverside, USA

Abstract

As core infrastructures housing the Internet and cloud services, data centers have been quickly expanding in both number and scale, resulting in a soaring power demand and collectively accounting for over 2% of global electricity usage. The rising energy price and heavy use of carbon-intensive electricity have undeniably placed an urgent emphasis on optimizing data center power management and building an energy-sustainable IT industry. Consequently, energy-efficient and green computing has emerged as a focal point of both computer systems research community and data center industry over the last decade, leading to impressive progress in energy efficiency of data centers as exemplified by the significant decrease in Power Usage Effectiveness (PUE) of state-of-the-art data centers from more than 2.0 to currently below 1.1. Nonetheless, best practices for energy efficiency and sustainability are currently largely restricted to a few mega-scale data center operators (e.g., Google and Facebook), and many challenges remain to be addressed to scale up the energy efficiency across the entire data center industry.

This course covers four broad topics in data center power management. First, we review the best practices for green data center design and operation, including both IT and non-IT resource management techniques. Second, we examine unique challenges and possible solutions to achieving sustainability in a very important type of data centers --- multi-tenant data centers, or commonly called “colocation” --- which have been largely hidden from the public and rarely discussed (at least in research papers). Third, we turn to the problem of reducing capital cost and maximizing utilization in data centers, and describe various techniques such as power/cooling oversubscription as well as their implications. Finally, we discuss the emerging data center demand response, which transforms the huge energy appetite of data centers from a negative into a highly valuable social asset.

This course is for students who have an interest in interdisciplinary research spanning one or more of the following topics: computer architecture, distributed systems, economics, and smart grid.

Bio

Shaolei Ren is an Assistant Professor of Electrical and Computer Engineering at University of California, Riverside. Previously, he was an Assistant Professor at Florida International University from 2012 to 2015. He received his B.E. from Tsinghua University in 2006, M.Phil. from Hong Kong University of Science and Technology in 2008, and Ph.D. from University of California, Los Angeles in 2012, all in electrical and computer engineering. His research interests include power-aware computing and systems, data centers, smart grid and network economics. He was a recipient of the NSF Faculty Early Career Development (CAREER) Award in 2015.


  Back to course info