IaaS下基于预测的弹性云服务的研究

针对供应虚拟机的IaaS （infrastructure as a service）云下虚拟机部署存在较高时延，导致弹性云服务效率低下的问题，提出一种基于ARIMA模型和季节指数的动态负载预测及资源估算的方法. 该方法利用负载与虚拟机配置的关系，预测负载值，并估算虚拟机需求量，从而可提前部署虚拟机，提高IaaS云的服务效率. 研究结合供应虚拟机的私有IaaS云环境，实现其下的弹性供应的资源决策模块. 实验与算法分析表明，该方法能够准确决策虚拟机资源量，保证虚拟机资源预留，有效改善了IaaS云的弹性效率.

Forecasting-based elastic cloud service research in an IaaS cloud

To deal with the issue that the delay time made by deploying virtual machine under the IaaS environment which provides virtual machine service for users leads low efficiency of elastic cloud service, an approach based on ARIMA model for predicting workload and forecasting the quantity of virtual machine in the future was proposed to settle the issue which has been leaded by time delay. Our method exploited the relationship between workload and the configuration of virtual machine resources, it predicted the workload firstly, and then estimated the quantity requirement of virtual machine. It improved efficiency of elastic cloud through reserving suitable virtual machine resource. Our research related to the private IaaS cloud environment which supplies virtual machine for users. The result of experiment and algorithm analysis revealed our method was accurate and practicable for decision of virtual machine.