The Optimized Energy Research For Virtual Machines In Cloud System

With the rapid development of cloud computing technology and the rapid expansion of cloud scale, the cloud deployment system requires more hardware facilities, which increases energy consumption and environmental load. The huge energy consumption of cloud computing has become the key problem of restricting its development. Nowadays, a large amount of computing resources and storage resources are concentrated in the cloud, which brings huge challenge for power consumption management of data center. So, the study about energy optimization in cloud computing system has become one of the most urgent research subjects.System energy consumption is mainly from IT system layer and infrastructure layer. Due to the existing shortcomings of the energy consumption and power consumption when measuring system overhead, this paper adopts the concept of energy efficiency as the benchmark of system modeling. The models of task load and energy consumption are built firstly, which is because that the two concepts are involved in modeling method based on energy efficiency, then the final model is reached from the ratio of task load and energy consumption. To calculate the value of energy efficiency, the model just needs physical host CPU usage and CPU frequency, which is convenient and effective.For virtual machine migratio n technology, this paper mainly studies three major problems in its technological process and optimizes their strategies to save energy consumption and improve energy efficiency:Firstly, the trigger strategy of migration should be chosen. The static double threshold method is adopted to determine whether the physical host is in low load or overload condition, and then, this paper put forwards the double threshold trigger strategy based on forecast. If the load of a physical host exceeds a certain threshold, its future load should be forecast to determine whether to start the migration. Secondly, the virtual machines which to be migrated are chosen. This paper analyzes the advantages and shortcomings of HPG and LPG, and the difference referred selection algorithm is proposed which reduces the amount of information and frequency of migration. Thirdly, it’s the choice of the target host. The available physical machine is whose CPU utilization between up threshold and low threshold. The hosts of meeting the conditions are chosen based on the comparison of host’s remaining resources and the current virtual machine’s required resources. This paper chooses the host during the above hosts whose idle resources is nearest the required resources as the target host.