The Dynamic VM Resource Scheduling By Using An Improved PSO Algorithm In The Cloud Data Center

Because of the cloud data center includes a variety of physical devices, network equipment and large amount of resource management, In many cases of a static resource management will appear some phenomenon that a few physical device under the heavy load and most physical device under the light load or idle. This phenomenon will lead to low utilization rate and cause unnecessary overhead. Therefore, achieving the goal of high efficiency and load balancing in Cloud data center is of great significance to solve the currently existing challenges.This paper studies and researches Cloud computing resource scheduling technology deeply and propose the VM deployment method based on the improved PSO algorithm. This paper mainly does the following work:(1)The article set up mathematical modeling abstracting the data center, quantifies CPU, memory and network, paying attention to utilization and load balance simultaneously with heuristic algorithm meanwhile analysis and description of resources in the data center are carried out.(2) Improving the PSO algorithm to fit the set of integers, and change the definition of the relevant parameter so that it can adapt to the mathematical model in the cloud data center. Improving the steps of the algorithm, Using Max-Min function to work out the optimal solution under the request of a variety of resources so that VMs are deployed in a better way, use load imbalance of the data center and multidimensional resource utilization of physical machine as optimization objectives to the algorithm reducing overhead in the data center. The algorithm increases the population quantity to expand the search areas so that there's less waste of resources in data center from a service provider's perspective.(3)Proposing the VM migration strategy based on the prediction. The paper designed the timing of the trigger migration in the process of scheduling dynamic virtual machine to determine when to adjust the system. Choosing the minimum number of migration strategy to select the appropriate virtual machine to migrate to the appropriate target physical machine. Choosing the best fit algorithm as the target physical machine selecting. Solving the following problems:1. When to migrate the VM 2. Which VMs are being chosen to be migrated? 3.which PMs are being chosen to be the target PMs.(4)The simulation experiments will be operated on the cloud emulator CloudSim. Compared with Round Robin algorithm based on greedy, PSO and improved PSO, we shall test all performance of virtual machine deployment and scheduling, and we compare the PMs CPU, memory, bandwidth utilization before the migration with the after migration scene to verify our algorithm effectiveness.