云环境下基于任务分类和LPM优化模型的调度算法

针对用户任务预算不足或期望完成时间较短，云服务方无法保障任务全部完成，提出基于任务分类和线性规划优化模型调度策略，使任务完成数最大化，同时考虑任务重要性。算法根据任务长短及重要性进行分类，然后建立任务计算资源关系矩阵及3个相关约束条件，以任务完成数最大化为目标函数，搭建线性规划模型，并给出算法实现。模拟实验表明，在同样的用户任务预算和期望完成时间条件下，该算法任务完成数最大值明显高于经典算法。

Scheduling Algorithm Based on Task Classifying and Linear Pogramming Model in a Cloud Environment

For smaller task budgets or higher demand for the time of task completion and the resource provider not providing all services, a scheduling algorithm based on linear programming model and task classification is proposed, with maximum number of performed tasks, and the priority of tasks is considered. The tasks are classified according to the priority and length of tasks. Then LPM is built with relation matrix and three constraints, taking the maximum number of performed tasks as objective function, and algorithms are given. The simulation shows that the maximum number of performed tasks is greater than that of classic algorithms with a smaller task budget or higher demand for task completion time.