先进计算基础设施中结点内部的任务调度和监控  

Task Scheduling and Monitor of Cluster in the Advanced Computational Infrastructure

作　　者：方兴[1] 李三立[1] 王小鸽[1] 黄震春[1] 何川[1] 

机构地区：[1]清华大学计算机系微机组,北京100084

出　　处：《计算机科学》2002年第6期13-15,23,共4页Computer Science

摘　　要：一、引言 在当前的高性能计算研究中,网络并行超级计算系统(集群式计算系统,又称cluster[1])是国内外研究的热点.网络并行超级计算可以以很低的价格提供高性能计算,能满足大多数用户的要求,特别适合在国内推广和使用.Advanced Computational Infrastructure (ACI) is a high-performance computational grid, which is based on shared high-performance computational resources and it is to solve the large-scale applications. The computational resources in ACT are dynamic, distributed and heterogeneous. This article introduces the architecture of ACI and emphasizes the task scheduel and monitor of cluster in ACI- We use XML to describe the resource in system, use a node server to control the cluster and use TCP/IP to communicate which is proceeded distributed. A schedule strategy and task priority are considerd to schedule tasks. The cluster and tasks are monitored when system is running.

关 键 词：先进计算基础设施 任务调度 监控 计算网格 异构操作系统 计算机 

分 类 号：TP316[自动化与计算机技术—计算机软件与理论] TP393[自动化与计算机技术—计算机科学与技术]