| Summary: | 博士 === 大同大學 === 資訊工程學系(所) === 101 === Cloud computing provides the highly scalable and maintainable computing resources for industrial applications and provides an important opportunity for information industry. In order to integrate the system resources, save energy consumption, and meet the user's requirements, Dynamic Voltage Frequency Scaling (DVFS) and virtual machine migration technologies were widely discussed for effectively reducing the energy consumption. Dynamic voltage and frequency scaling is to use the system slack time to reduce the system voltage, so as to reduce energy consumption, but the execution time of the work will be prolonged. It is an important issue to effectively adjust the operating voltage of each job, and to meet the users’ requirements. In recent years, many researchers proposed a number of different scheduling algorithms based on dynamic voltage frequency scaling technique to solve this problem, but most of those studies applied static scheduling methods which limited to the known execution order of tasks. A dynamic scheduling mechanism in the cloud computing environment, called Extenics-based Dynamic Voltage Frequency Scaling (E-DVFS) mechanism has been proposed in this dissertation. In the cloud computing environment, applying virtualization technologies, according to the expected execution time provided by the user, an extenics relational function can be built to calculate and predict the completion time of the job and the working voltage. Moreover, the dynamic virtual machine migration technique has been applied to further reduce energy consumption and meet the Quality-Assurance (QA) requirements. CloudSim simulator is used in this research to verify the effectiveness of the proposed E-DVFS mechanism. Simulation results show that the system can meet QA requirements, and total energy consumption of servers can be reduced about 13% ~ 25 %.
|
|---|
