The Virtual Machine Placement Algorithm Based on Equivalent Resource Model in the Offloading Data Center of Mobile Cloud Environment

摘要

In the mobile cloud environment, the massive computing and interchanging tasks are necessary to split and migrate from the mobile terminals to the matched virtual machines in offloading data center. In general, the constant and intensive resource application and virtual machines placement actions probably cause resource fragmentation phenomenon in the physical Servers of data center. The performance overhead seriously constraints the number of parallel load virtual machines and thereby obstructs the global efficiency of offloading system. In this paper, we proposed a novel online virtual machine placement algorithm. Primary, it establishes a virtual machine equivalent peak resource utilization model for one-dimension resource constrained condition, which is on account of historical static data, law of large numbers and given Service Level Agreement(SLA). Secondary, according to the equivalent model, it is able to simplify and convert the placement process to a improved bin packing algorithm via SLA constrained and virtual machine random exchange strategy. Tertiary, based on polling strategy and statistic theory, it builds placement arrangement algorithm for multi-dimension resource condition. Theoretical analysis and numerical simulation demonstrates that the online algorithm is balanceable of the efficient and fairness for virtual machines placement in resource reservation and allocation of one-dimension and multi-dimension condition.