Author Login Editor-in-Chief Peer Review Editor Work Office Work

Computer Engineering ›› 2007, Vol. 33 ›› Issue (17): 99-101. doi: 10.3969/j.issn.1000-3428.2007.17.034

• Software Technology and Database • Previous Articles     Next Articles

Large-scale Job Management System Based on Partition and Lease Technologies

ZOU Ming1,2, TU Bi-bo1,2, ZHAN Jian-feng1   

  1. (1. Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100080; 2. Graduate School, Chinese Academy of Sciences, Beijing 100039)
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-09-05 Published:2007-09-05

基于分区租借的大规模作业管理系统

邹 铭1,2,涂碧波1,2,詹剑锋1   

  1. (1. 中国科学院计算技术研究所,北京100080;2. 中国科学院研究生院,北京 100039)

Abstract: Partitioned workload system (PWS) is one job management system using partition and dynamic resources-lease technology, which enhances the manageability of JMS and improves the utilization of cluster system. With optimized communication system, PWS can support cluster systems with thousands of CPUs or nodes. PWS has been applied in Shuguang 4000A super server and other computation fields requiring high performances. the design and implementation of the PWS is introduced.

Key words: JMS, cluster system, partition, lease, scalability

摘要: PWS(partitioned workload system)是引入分区化技术以及资源租借技术的作业管理系统(JMS)。分区技术加强了机群系统中资源的可管理性,通过资源租借技术提高了系统资源的利用率。优化的通信方式让PWS能够支持上千节点的大规模机群系统。PWS已经在曙光4000A超级计算机以及相关高性能计算领域中得到了应用。该文介绍了PWS的设计与实现。

关键词: JMS, 机群系统, 分区, 租借, 扩展性

CLC Number: