云计算平台的自适应资源供给

Adaptive Resource Provisioning for Cloud Computing

云计算系统中IaaS层通过对虚拟化后的基础设施进行池化管理来完成云内基础设施资源的管理。资源的池化为云服务提供按需索取的资源供应以及动态的资源配给。针对现有云计算平台缺乏对资源供应量的自动调整机制的问题,研究了云计算平台中基础设施资源供给的自适应性。通过二次平均时间序列预测法对未来一个时段内的业务负载峰值进行预测,并将预测值交予云平台转化为资源需求。对给定的资源需求,模型通过不断寻找最小虚拟机所能提供的资源与预期资源需求量之差的向量长度,做出虚拟机调度决策。仿真实验表明,本文提出的云计算自适应模型具有良好的精确性及稳定性。

An infrastructure cloud manages the infrastructure as a resource pool, enabling on-demand and dynamic resources provisioning according to resource consumption and service loads. This paper investigates the issue of adaptive resource provisioning for services in clouds. A model is proposed to predict cloud services＇peak load and calculate the expected resources consumption. Resource of the best utility for the service is identified and allocated to the service. Simulation results suggest that the proposed adaptive resources provisioning model has a fair accuracy and stable performance.