基于层次化调度策略和动态数据复制的网格调度方法

针对在网格中如何有效地进行任务调度和数据复制, 以便减少任务执行时间等问题, 提出了任务调度算法（ISS）和优化动态数据复制算法（ODHRA）, 并构建一个方案将两种算法进行了有效结合。该方案采用ISS算法综合考虑任务等待队列的数量、任务需求数据的位置和站点的计算容量, 采用网络结构分级调度的方式, 配以适当的权重系数计算综合任务成本, 搜索出最佳计算节点区域; 采用ODHRA算法分析数据传输时间、存储访问延迟、等待在存储队列中的副本请求和节点间的距离, 在众多的副本中选取出最佳副本位置, 再结合副本放置和副本管理, 从而降低了文件访问时间。仿真结果表明, 提出的方案在平均任务执行时间方面, 与其他算法相比表现出了更好的性能。

Data replication and scheduling algorithm in data grids

This paper proposed the job scheduling algorithm (ISS) and optimization of dynamic data replication algorithm (ODHRA) for effectively conducting job scheduling and data replication, and built a program to effectively combine the two algorithms, in order to reduce the execution time of job. First, this program used ISS algorithm to give a comprehensive consideration to the number of jobs waiting in queue, the location of required data for the job and the computing capacity of sites, and used a hierarchical scheduling of network structure with appropriate weighting coefficient to calculate the comprehensive job cost and search for an optimum computing node area; next, it used ODHRA algorithm to analyze data transfer time, storage access latency as well as replica requests that waiting in the storage queue and the distance between nodes, selected the best replica location from many replicas, and then combined replica placement and replica management to reduce the file access time. The simulation results demonstrate that the proposed replication and scheduling strategies give better performance than the other algorithms in the mean execution time of job.