الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 133 |  |  |
| | | from | 133 | | |
Abstract
Resource utilization and minimization of carbon dioxide (CO2) are the main factors in cloud data centers that should be taken into consideration. According to live migrations of VMs, researchers proposed different techniques to optimize VM migration process, including the selection of most appropriate VMs to be migrated from the overloaded hosts to other available hosts and choosing the most appropriate host to allocate these migratable VMs. The main objective of this thesis is to improve the process of virtual machines migration in order to achieve load balancing between physical servers in datacenter, which leads to improve the resource utilization without violation in the Service Level Agreements (SLA) that established between the end users and cloud service provider.
Four new algorithms are proposed; the first algorithm called CPU Utilization Variance (CUV) is based on maximizing and balancing CPU utilization for all servers. The second and the third algorithms are based on memory utilization of virtual machines (VMs) that are migrated from the overutilized host to other hosts. In the second algorithm called Minimum Memory Utilization (MNMU), the best VM to be migrated is the VM with the minimum memory utilization. In the third technique called Maximum Memory Utilization (MXMU), the best VM to be migrated is the VM with the maximum memory utilization. The fourth algorithm called Minimum Memory CPU Utilization Variance (MMCUV) is based CUV and MNMU algorithms. It depends on the CPU utilization variance between all hosts and also it depends on the memory utilization of the migrated VMs. Further, there is a comparative study performed between the proposed algorithms and other recently algorithms. The experiments have been implemented by using CloudSim toolkit.
The comparative study has been performed between the proposed algorithms and most resent algorithms such as Maximum Correlation (MC), Minimum Migration Time (MMT), Minimum Utilization (MU) and Random selection (RS).
Energy consumption in KWH, performance in the form of Million Instructions per Second (MIPS), number of virtual machine migrations and Service Level Agreement Violation (SLAV) metrics are used for evaluating the different algorithms. The results of the experiments show that the lowest energy consumption without violating SLA on a large-scale data centers is achieved by the proposed techniques which obtains higher efficiency for maximizing resources utilization and minimizing carbon dioxide (CO2) emissions. CO2 emissions are increased due to energy consumption in the data center. The aim of this thesis is to reduce the energy consumption in the data center that leads to reducing in CO2 emissions.