摘要
根据电网数据特征,提出基于树形结构分布式对象缓存的任务调度算法。算法将业务数据封装成数据对象,将工作任务抽象成任务对象,按照树形结构将数据对象和任务对象分布到集群计算机各节点,实现对象分布式的并行计算。设计了对象化并行计算框架并选取电网可靠性计算典型业务场景,对算法进行了验证和应用。算法对大幅提升数据型电网统计分析的计算效率具有重要作用。
According to the characteristics of grid data, this paper proposes a task scheduling algorithm using the distributed object cache based on the tree structure. The algorithm packages business data into a data object, abstracts the task into a task object, and then distributes data objects and task objects into different computing nodes through the tree structure ill order to achieve the goal of object-oriented distributed parallel computing. This paper designs an object-oriented distributed parallel computing, and verifies and applies the algorithm in a typical business scene for grid reliability calculation. The algorithm plays an important role in greatly improving the computational efficiency of data grid statistics and analysis.
出处
《电力信息与通信技术》
2014年第4期19-24,共6页
Electric Power Information and Communication Technology
关键词
调度方法
树形结构
分布式计算
对象化并行计算框架
scheduling algorithm
tree structure
distributed computing
object-oriented parallel computing framework
