期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2014
卷号：63
期号：2
出版社：Journal of Theoretical and Applied
摘要：Job scheduling plays an important role in cloud computing networks. Appropriately distributing service jobs to multiple servers when the cloud network stays in dynamic load conditions is the main work of the scheduling strategy. We adopt the resource load level of a server as the state of the server, and convert the job processing in a server to a semi-markov process. A job scheduling strategy can be given by the optimization algorithm to make the running cost minimum and improve the resource utilization. A simple job scheduling strategy based on resource load balance is given first in this study. Then, the coefficient vector which indicates the weight of the various idle resources of the simple strategy is extended to a coefficient matrix which can meet the dynamic resource load conditions. An optimization algorithm with the coefficient matrix is adopted to reduce the running cost by iterations. In the following sections, we improve the optimization algorithm by considering the shared resources and studying the best division of the resource load levels. At the last of the iteration, a proof of the convergence of the iteration is given. Finally, simulations and numeral results are provided.
