Dynamic memory balancing for virtual machines

摘要: Virtualization essentially enables multiple operating systems and applications to run on one physical computer by multiplexing hardware resources. A key motivation for applying virtualization is improve resource utilization while maintaining reasonable quality of service. However, such a goal cannot be achieved without efficient management. Though most resources, as processor cores I/O devices, are shared among virtual machines using time slicing can scheduled flexibly based priority, allocating an appropriate amount main memory more challenging. Different have different requirements. Even single application shows varied working set sizes during its execution. An optimal management strategy under virtualized environment thus needs dynamically adjust allocation each machine, which further requires prediction model that forecasts host the fly. This paper introduces MEmory Balancer (MEB) monitors usage accurately predicts needs, periodically reallocates memory. MEB uses two effective predictors which, respectively, estimate available reclaiming notable performance drop, additional required reducing machine paging penalty. Our experimental results show our schemes yield high accuracy low overhead. Furthermore, overall system throughput significantly improved with MEB.