Title :
Energy and locational workload management in data centers
Author :
Spatari, Sabrina ; Kandasamy, Nagarajan ; Kusic, Dara ; Ellis, Eugenia V.
Author_Institution :
Dept. of Civil, Archit., & Environ. Eng., Drexel Univ., Philadelphia, PA, USA
Abstract :
Data centers are growing consumers of energy and emitters of greenhouse gases (GHG) worldwide. This paper examines data center power and locational workload management as strategies for energy savings and GHG emissions reduction. The case study examined focuses on GHG emissions from the electricity grid supply in a controlled small-scale computer cluster experiment and location (Philadelphia, PA). Virtualization is a technique that consolidates multiple online services onto fewer computing resources within a data center and deploys computing resources only as needed. The method can be applied not only within a data center, but also among multiple data centers in different locations, thereby taking advantage of deploying data centers that are linked to “low-carbon” electricity grids. Understanding the interaction between data center location and real time power consumption is critical to optimizing computer cluster usage, since demand during certain times of the day may rely on coal as the marginal source. Using the power savings results generated from the virtualization experiments performed on a small computer cluster at Drexel University, and power supply from the electricity grid serving the data center over a 24-hour day during a peak electricity summer month, we examine the time of day for shifting data center workloads in order to minimize GHG emissions.
Keywords :
air pollution control; computer centres; power aware computing; power consumption; power grids; GHG emission reduction; controlled small-scale computer cluster experiment; data center; electricity grid supply; greenhouse gase; locational workload management; marginal source; power consumption; power savings; power supply; Buildings; Coal; Cooling; Data models; Electricity; Power grids; Servers; Energy; Information technology; Life cycle assessment; Power management;
Conference_Titel :
Sustainable Systems and Technology (ISSST), 2011 IEEE International Symposium on
Conference_Location :
Chicago, IL
Print_ISBN :
978-1-61284-394-0
DOI :
10.1109/ISSST.2011.5936862