Dynamic Rightsizing with Quality-Controlled Algorithms in Virtualization Environments

Abstract

Virtualization and partitioning are the means by which multiple application instances can share and run multiple virtual machines supported by a platform. In a Green Cloud environment, the goal is to consolidate multiple applications onto virtual machines associated by fewer servers, and reduce cost and complexity, increase agility, and lower power and cooling costs. To make Cloud center greener, it is beneficial to limit the amount of active servers to minimize energy consumption. This paper presents a precise model to formulate the right-sizing and energy-saving mechanism, which not only minimizes energy consumption of the server but also maintains a service quality through the Mt/M/Vt strategy of queuing theory. The authors map the complicated formula of the energy-saving mechanism to an approximation equation and design the fast decidable algorithms for calculating the right size of virtual machines in constant time complexity for power management systems.