论文标题
认知云数据中心的任务录取控制和边界分析
Task Admission Control and Boundary Analysis of Cognitive Cloud Data Centers
论文作者
论文摘要
与批处理任务相比，这里研究了一个新颖的云数据中心（DC）模型，具有实时（或在线）流的认知能力。在这里，DC可以确定使用资源的成本，在线用户或具有批处理任务的用户可能会决定是否为获得服务付费。在线服务任务在获得服务而不是批处理任务方面具有更高的优先级。两种类型的任务都需要一定数量的虚拟机（VM）。通过定位总折扣奖励的最大化，最终确认了承认任务任务的最佳政策是与国家相关的控制限制策略。接下来，用于现实中的估计和利用率分别得出了这种最佳政策的下部和上限。最后，在各种情况下进行了一组综合实验，以验证该提出的模型和解决方案。作为演示，采用了机器学习方法来展示如何通过使用前馈神经网络模型获得最佳值。本文获得的结果将被预期用于具有经济最佳策略的各种云数据中心。
A novel cloud data center (DC) model is studied here with cognitive capabilities for real-time (or online) flow compared to the batch tasks. Here, a DC can determine the cost of using resources and an online user or the user with batch tasks may decide whether or not to pay for getting the services. The online service tasks have a higher priority in getting the service over batch tasks. Both types of tasks need a certain number of virtual machines (VM). By targeting on the maximization of total discounted reward, an optimal policy for admitting task tasks is finally verified to be a state-related control limit policy. Next, a lower and an upper bound for such an optimal policy are derived, respectively, for the estimation and utilization in reality. Finally, a comprehensive set of experiments on the various cases to validate this proposed model and the solution is conducted. As a demonstration, the machine learning method is adopted to show how to obtain the optimal values by using a feed-forward neural network model. The results achieved in this paper will be expectedly utilized in various cloud data centers with cognitive characteristics in an economically optimal strategy.
