| As an important component of future network, Internet of things(Io T) and cloud computing have been applied to many aspects of our daily life. However, with the continuous expansion of the size of Io T and growing number of connected terminals, how to deal with the mass data by Io T terminal with limited calculation ability and storage capacity has been a great challenge. To tackle this challenge, cloud computing has been proposed as a promising solution due to its powerful computing and storage capacity. The combination of Io T and cloud computing technology has become a research hotspot.This thesis first analyzes the necessity of the combination of Io T and cloud computing and presents a data service platform based on Io T demand, which acts as middle layer of the Io T architecture. The virtual resource scheduling of the data service platform is one of the issues that must be considered and is also the the main contents of this thesis.Next, for the issue of virtual machine mapping, this thesis proposes a multi-cube mapping algorithm, in which virtual resource is characterized by vectors and the total resource of data resource platform is characterized by a multi-cube. We adopt the model of three dimensional vector packing and use the criterion of complementary resources and minimum unbalance degree to complish the virtual machine mapping. Simulation results confirm the obvious advantages of the proposed multi-cube mapping algorithm in minimizing server usage, improving resource utilization, and reducing the energy consumption of data center.Finally, for the issue of virtual machine migration, this thesis devises a migration algorithm based on minimum load balance, in which we adopt the model of multi-objective optimization, set resource imbalance as objective function, and take multi-objective genetic algorithm to obtain optimum solution. Simulation results show that the proposed migration algorithm can achieve load balance effectively and reduce the cost of migration. |
PDF Full Text Request