ABSTRACT
Today’s “Internet-scale” systems may be made up of several hundred or thousand servers spread across many geographies. ese systems consume several megawatts of electricity a day. It is important therefore to build systems that are optimized for power management. However, building such a system is a challenge as trade-offs between application performance and power consumption need to be considered. In this chapter, we discuss recent advancements in cloud computing and smart grid technologies to design a power management system that helps reduce the power expenditure incurred by a cloud provider without “overtly” sacrificing the performance of the applications hosted by it. In particular, this chapter discusses ways in which a cloud provider can respond to various dynamic pricing signals received by the smart meters installed at its facilities, called data centers, by autonomously moving “noncritical” applications to remote sites during peak electric grid load situations by leveraging techniques from cloud computing.
