Energy-efficient Online Scheduling Algorithm With The Life Cycle Consideration Of Virtual Machines In Cloud Data Centers

The success of cloud computing has led to the establishment of large-scale data centers to meet the growing computing power on demand, but the data centers consume huge electrical energy, how to manage the resources in the cloud data center efficiently is a great of challenge. Cloud data centers consume large amounts of energy, increased carbon dioxide emissions, and brought serious pollution to the environment. So the reasonable design of energy-efficient scheduling algorithm to manage the cloud data center resources is great of significance, which can not only reduce the energy consumption of data centers, but also reduce the operation cost of enterprises.Considering many factors for energy-efficient scheduling algorithm is NP-Hard in general. For most of the task scheduling, the key factors such as the life cycle of virtual machines(VMs) and the total running time of physical machines(PMs) are not be considered for energy consumption. One best known for static offline energy-efficient scheduling algorithm is a 3-approximation MFFDE algorithm, while for dynamic online energy-efficient scheduling research is relatively less, the most famous is the g- competitive GREEDYBUCKET algorithm, g is the total CPU capacity of a physical machine, g ?2.This thesis considers online energy-efficient scheduling in cloud data center. The goal of online energy-efficient scheduling of virtual machines is non-preemptively in their start time to end time windows, minimizing the total busy time of all used PMs. This thesis proposed a ? ?competitive online GRID algorithm and a 22 1(1)g gcompetitive k k? ?? ? ? BFF algorithm, where 1?? ?g and k is the ratio of the longest interval over the second longest interval for k ?1. In order to further improve the utilization of data center resources, combining VM migration, we proposed the GRID migration algorithm and BFF migration algorithm. Through intensive simulation tests, we found that the GRID and BFF algorithm can save energy on average 10%-20%, and GRID migration and BFF migration can further reduce the total energy consumption by 5%- 10%.