Research And Implementation Of Virtual Server Consolidation Based On Forecasting In Cloud Environment

In cloud environment, the virtual server consolidation is to use virtual machines to encapsulate applications running on multiple physical servers and then integrate them into a small number of servers. The consolidation technology can reduce large number of servers used, and also improve server utilization greatly. Therefore, it has been the commonly used method in enterprises to reduce energy consumption, hardware and operating costs. The consolidation technology becomes a hot issue in cloud computing. A virtual server consolidation strategy based on prediction is studied and implemented. The major work includes:(1) The current state of research on cloud computing and virtual server consolidation of cloud computing are analyzed. Related technologies of virtualization are summarized. And then the technology of virtual server consolidation is analyzed, including consolidating model, consolidating features, consolidating targets and consolidating algorithms. The current state of research on virtual server consolidation algorithm is introduced and the shortcomings are analyzed.(2) To solve the problems that the load forecasting models of cloud computing take insufficient account of the correlation between different loads and have a low accuracy, a load prediction algorithm based on improved cloud model is proposed. The proposed method improves the prediction accuracy by considering the correlation between CPU and memory. By presenting the concepts of similar cloud and overlap cloud to obtain the zooming conceptions cloud model, the load forecasting model based on the improved cloud model is established. The simulation results show that the forecasting model based on improved cloud model has advantages than the other forecasting models on maximum relative error, minimum relative error and average error, and the presented algorithm can provide higher forcasting accuracy.(3) Aimed at how to consolidate a large number of virtual machines to a physical server, this thesis puts forward a virtual machine consolidation strategy based on cuckoo search algorithm. Firstly, the virtual machine consolidation problem is abstracted into an optimization model. The optimization model considers the energy consumption of virtual machine migration and network equipment, and introduces the prediction mechanism to avoid unnecessary secondary migration. Secondly, considering the shortcomings of slow convergence speed and low convergence precision owing to the lack of effective coordination mechanism in the later period of cuckoo search algorithm, the thesis improves the traditional cuckoo algorithm by setting the detection probability and step size dynamically and introducing disturbing factor, and applies it to solve virtual machine consolidation problem. The results of simulation show that the virtual machine consolidation strategy based on the cuckoo algorithm has better performance on the aspects of the number of servers, the times of virtual machine migration and consolidation energy.(4) A simulative consolidation system of virtual server based on prediction is designed and implemented in cloud environment. The system extends and designs the corresponding function modules of CloudSim(a simulation platform for cloud), which includes environment configuration module, load forecasting module and virtual sever consolidation module. Finally, the effectiveness of the proposed system is verified by testing the system.