论文部分内容阅读
云计算环境为用户提供了大量弹性可扩展的基础设施资源,用户可以按需购买和支付服务.如果云平台能够自动分配给一批任务合理规模的计算资源,将会方便用户使用并且较大节省用户服务成本.本文在瀚海星云云计算平台上构建了一个软件应用服务层,提供Bag-of-tasks(BoT)应用公共服务.根据历史信息通过回归方法和BP神经网络方法对BoT中的子任务进行执行时间预测.然后使用最大并发度概念,在虚拟机内存是否满足任务情况下,提出了VMA,NP-IO和NP-DP三种算法.最后,使用图片分割软件作为BoT应用,从资源分配情况、任务完成率和算法时间复杂度方面验证了算法的有效性.
Cloud computing environment provides users with a large number of flexible and scalable infrastructure resources, users can purchase and pay for services on demand.If the cloud platform can automatically assigned to a number of tasks of a reasonable scale of computing resources, will be user-friendly and greater savings User service cost.In this paper, a software application service layer is built on the Hanhai cloud cloud computing platform to provide Bag-of-tasks (BoT) application public service.According to the historical information, the regression of sub-tasks in BoT And then use the concept of maximum concurrency to propose three algorithms of VMA, NP-IO and NP-DP when the virtual machine’s memory meets the tasks.Finally, using image segmentation software as BoT application, Situation, task completion rate and algorithm time complexity verify the effectiveness of the algorithm.