论文部分内容阅读
层次存储技术是存储和管理海量连续媒体数据的有效手段。I/O请求的调度是层次存储研究中的重要问题,目前也有很多的调度算法,但它们之间缺乏定量的分析和对比,使得调度算法的选择缺少可靠的依据。该文利用排队论对调度收益进行定量分析,分析了它与系统平均队长等因素的定量关系,并得到调度在平均等待时间上的收益与原始等待时间的关系,指出它们近似成二次曲线关系。仿真试验表明了这些结论的有效性,它对于层次存储管理系统中,I/O请求调度算法的选择具有很好的指导作用。
Hierarchical storage technology is an effective means of storing and managing massive amounts of continuous media data. Scheduling of I / O requests is an important issue in Hierarchical Storage. Currently, there are many scheduling algorithms, but the lack of quantitative analysis and comparison between them makes the selection of scheduling algorithms lack of reliable basis. This paper makes use of queuing theory to quantitatively analyze dispatch revenue, analyzes the quantitative relationship between it and system average captain and other factors, and obtains the relationship between the dispatch’s return on average waiting time and the original waiting time, pointing out that they approximate to a quadratic curve . Simulation results show the validity of these conclusions. It has a good guidance for the selection of I / O request scheduling algorithm in Hierarchical Storage Management System.