The Status Prediction of Physical Machine in IaaS Cloud Environment

Abstract

At present, in researches of Iaas cloud resource scheduling strategies, it is focused that SLA violation or overloaded physical machine can trigger the migration of virtual machines, which will reduce the performance of the system and cause extra energy cost. In this paper, we model the resource of IaaS cloud based on Hidden Markov process to predict the status and the time that the physical machine is overloading, which will serve as a guideline for the resource scheduling in the IaaS cloud. Specifically, the resource status of physical machine will be chosen as the hidden status, meanwhile, the operations of virtual machine will be an observation set of the visible status, which are a modelling process. And then, we present the optimal path of the status transition probability as the core method of the physical machine status prediction. Finally, through real experimental scenarios, we verify the effectiveness of physical machine status prediction in the IaaS cloud environment.