Energy efficient power aware VM scheduling in decentralized cloud

摘要

The increase in the popularity of cloud computing paradigm in the recent years has led to the rapid expansion of datacenters which involves deploying more number of software networks and remote servers that allow centralized storage of data and online access to the services or its resources. In spite of its advantage of reduced energy consumption at the user's end, managing the increase in demand, optimal energy efficiency and satisfying high performance expectations of users must be taken care at the cloud provider's end. In this paper, we propose a decentralized Power Aware Best Fit Priority algorithm that allocates the requested virtual machines to the hosts whose increase in the power consumption factor is minimal. We also handle the host overloading and underloading scenarios by including the concept of “skewness” to detect the uneven utilization of servers and by adopting Median Absolute Deviation(MAD) policy to dynamically set the upper and lower thresholds for VM consolidation.