The Study of Key Technology for Dynamic Resource Management in the Cloud Computing Environments

Storage resource scheduling is lack of flexibility, and user’ demand for different data is vary. In cloud computing storage resource scheduling process, these data files to be equally are irrational, so it needs a dynamic storage resource scheduling mechanism to distinguish requirements of different data. For the file hot issues that a surge request results in the copy number of the current file is difficult to meet the demand of Cloud computing, we presented a cloud storage resources dynamic scheduling mechanism based on Hadoop distributed file system and genetic algorithm. We used tree structure to encode and decode chromosomes, used the spanning tree algorithm to generate the initial population of virtual machines. We designed the corresponding fitness function evaluation program. And according to the changes in chromosome structure we designed selection operator, crossover operator and mutation operator. Experimental results show that the algorithm compared with traditional algorithms, convergence time is faster, while effectively reducing the number of virtual machine migration.