面向任务服务质量的网络资源需求预测算法

A Resource Requirement Prediction Algorithm for Task-Oriented Quality of Service

在信息化时代背景下,云计算作为一种全新的共享基础资源架构的模式,给予用户更加经济和便利的解决方案。一般情况下,用户对于完成任务所需求资源量是不了解的,用户只能大概估计需要的资源需求量。这样不仅会承担不必要费用,也会造成云资源浪费。为了满足不同用户对不同任务需求,并优化资源分配,论文提出了一个面向任务服务质量的资源需求预测方法,用户仅需提出任务的Qos指标,算法自动预测能够保障用户最小的网络资源供给量。在此基础上,调度器根据预测结果实现用户需求与网络资源之间的匹配。实验表明该方法可精确预测用户需求,并且能提高网络资源的利用率。

In the background of the information age,cloud computing as a new model of sharing basic resource architecture gives users more economical and convenient solutions.In general,user does not know the amount of resources required to complete tasks,and can only estimate the required resource demand approximately.In this condition,the user not only bears unnecessary costs,but also wastes cloud resources.Focusing on meeting the requirements of different users for all kinds of tasks and optimizing resource allocation,this paper proposes a resource demand forecasting method.The user only needs to submit the Qos parameters of the task,and the algorithm automatically predicts the minimum resource supply to guarantee the Qos demand.The experimental re sults show that this method can accurately predict user requirements and improve resource utilization.