Research On Approach Of Measurement And Prediction Of Performance Interference Between Virtual Machines And Its Application

Cloud computing is a kind of new service delivery and usage pattern, it aims to provide users with safe, fast and convenient data storage and computing services. Meanwhile, the dynamic expansion and pay-on-demand and other features make it well positioned to meet the demand of control service costs and reduce energy, so it has a broad market prospect.In the cloud computing data center, virtual machine is the minimum size of resource management and allocation, which users usually want to be able to run stably, so how to effectively implement allocation of virtual machine has become a research hotspot. However, most of the current research ignored the influence of performance interference between virtual machines, and can not guarantee the validity of the virtual machine placement.In order to guarantee the validity of resource allocation scheme in cloud computing, this paper studied related work about resources allocation and metric of performance interference between virtual machines, proposed an approach of measurement and prediction of performance interference between virtual machines in cloud environment. This approach has three steps:First, build a PID model and calculate the model parameters. Second, use the hierarchical clustering algorithm for virtual machine load patterns mining. Third, through matching the current virtual machine load patterns with clustering results, completing the PID prediction. Finally, through a series of virtual machine placement experiments validate that the proposed PIMP can guarantee the QoS effectively.