Energy-saving Algorithm Research Of Computing Resources In Data Center

At present, there are prevalence of high energy consumption, serious waste ofresources and other issues in data center. How to efficiently integrate computingresources and reduce energy consumption has become a research hotspot in data center.Because of the advantage of efficient use of resources and simple management withadopting virtualization technology, data center can achieve the goal of load balancing ofresources and reducing energy consumption with adopting energy-saving algorithmbased on migration of virtual machine to allocate and schedule resources. However,most of energy-saving algorithms did not consider energy consumption of migration ofvirtual machine between servers, so an algorithm that excessively depends on themigration of virtual machine to achieve the average distribution of resources may notachieve the goal of reducing the energy consumption of the system.Therefore, studying the energy consumption during the migration of virtualmachine and designing a reasonable energy-efficient scheduling algorithm for energymanagement in data center have a very important significance. Specifically, the thesis’smain contents and achievements include.1. Model power consumption of server in data center. This thesis studies the factors,such as CPU utilization, memory utilization and the case of disk reading and writing,which affect the energy consumption of the server, and combines experimental data toestablish the mathematical model of energy consumption.2. Model power consumption of migration of virtual machine. This thesis studiesthe factors of affecting migration performance of virtual machine, performance loss ofvirtual machine and energy consumption of the server during the migration of virtualmachine, and establishes the model of energy consumption of migration of virtualmachine.3. Design Energy-saving algorithm. Based on the study of consumption model ofserver and the consumption model of migration of virtual machine, this thesis designsan energy-saving algorithm of largest span of offline load and an online algorithm ofmigration of virtual machine. The former considers the span of the load range to achieve reasonable allocation of resources and the shortest turn-on time of server. The latterconsiders the migration of virtual machine to satisfy user requests with the smallestservers, and finally achieves to reduce the energy consumption of data center.4. Verify energy consumption of migration of virtual machine. Through thecollection and analysis of energy consumption data of server during the migration ofvirtual machine, we have verified the energy consumption overhead that is present inthe process of virtual machine migration. The experimental results show CPU utilization,VM sizes of memory and network bandwidth have a greater impact on energyconsumption of the server during the migration of virtual machine.5. Contrast the performance of different algorithms. By comparing with LoadBalancing algorithm, Dynamic Round Robin algorithm, Energy-aware Migrationalgorithm and Modified Best Fit Decreasing algorithm, the experimental results showthat the energy-saving algorithm of largest span of offline load and the online algorithmof virtual machine migration have obvious advantages in total energy consumption ofthe data center, turn-on time of the server, running time of the algorithm and so on.