Large data center operators such as Google, eBay, Microsoft and others have the ability and resources to optimize all aspects of their data center, including reliability, availability and energy efficiency. These operators are able to unify control and management aspects at all levels and boundaries in the data center starting with customized server designs to cooling systems to sensing, overall control, and power distribution and to the service classes offered. However, there is a large percentage of data centers that use commercial, off-the-shelf (COTS) components for their data center and come up with specific management techniques that may not be at the leading edge. This panel discussion focuses on emerging techniques for bringing these COTS-implemented data centers up to the agility, reliability and efficiency levels that represent 21st-century practices. The panel also aims to initiate a discussion on enabling technologies, design and operational guidelines for COTS-based data centers. The panelists will have short presentations and will be asked to address the pros and cons of requiring data centers to be run by intrinsically intelligent operating systems utilizing controls, distributed sensing and self-learning algorithms.
Moderator:
Bahgat Sammakia (bio)
Center Director, Center for Energy-Smart Electronic Systems (ES2) and
Vice-President for Research, Binghamton University
Panelists:
Dereje Agonafer (bio)
Jenkins Garrett Professor
Site Director, NSF IUCRC Center in Energy Efficient Systems
Director, Electronics, MEMS & Nanoelectronics Systems Packaging Center
University of Texas, Arlington
Kanad Ghose (bio)
BU Site Director, ES2 and
Professor of Computer Science, Binghamton University
Roger Schmidt (bio)
IBM Fellow Emeritus, Traugott Distinguished Professor,
Mechanical and Aerospace Engineering Dept.,
(Retired IBM Chief Engineer for Data Center Energy Efficiency) Syracuse University
Mark Seymour (bio)
CTO
Future Facilities