【作者】 胡威；

【导师】 吴克河；

【作者基本信息】 华北电力大学（北京） ， 计算机应用技术， 2008， 硕士

【摘要】 本文提出的任务调度算法是针对实时数据平台中传统任务调度存在的负载不均衡及任务实时性差等问题给出的解决方案。在查阅了大量的国内外文献的基础上,对实时调度及负载均衡技术的基础理论和主要方法进行了深入的学习和研究。针对电力系统实时数据平台中实时任务的特点,提出了一种基于优先级的动态反馈的调度算法。该算法既考虑了服务节点的负载率,又兼顾了实时应用系统的一个重要指标——截止期错失率,充分保证了负载的均衡性和任务的实时性,从而大大改善了实时数据平台的性能。最后,本文为实时数据平台设计和实现了一个任务调度服务,并通过实验对其性能进行了分析,验证了调度方案的优越性。更多还原

【Abstract】 The task scheduling algorithm this paper proposed is a solution which aimed to solve the problems of imbalanced load and poor timeliness with the tradition task scheduling in the real-time data platform. Based on investigating a great deal of domestic and international documents, the basic theory and main method of real-time scheduling technology and load balancing technology is researched deeply. In view of the characteristic of task in real-time data platform for power gird, a task scheduling algorithm based on the priority and dynamic feedback is proposed. This algorithm both has considered the rate of loading on server nodes, and has given attention to an important index——Deadline Miss Ratio. The algorithm guaranteed the loading balance and timeliness of task fully, and improved the performance of real-time data platform greatly. Finally, this article designed and implemented a task scheduling service for the real-time data platform, and analyzes the performance through the experiment.更多还原