一种网格系统的网络管理系统的设计与实现

Design and Implementation of Network Management Server for Grid Systems

网格计算的一个显著的特点是多系统相互协作,共同解决一个复杂问题或者实现复杂的功能。分析了网格计算给NMS(网络管理系统)带来的挑战,以及主流网络管理技术(包括SNMP,CI-MP,CORBA,LDAP,TMN等)的缺陷,提出了一种基于改进的JMX和OpenNMS的针对网格分布式协作系统的NMS设计和实现。该实现的重要特点是能够快速、小代价地改造现有服务器,快速接入NMS,还可以精确定位分布式协作系统的故障。

Grid Computing has a key characteristic, that is, multiple systems cooperation and solving a complex program or realizing complex function together. This paper starts with the challenge of NMS design brought by grid computing, then analyzes the defects of the mainstream network management techniques （ including SNMP, CIMP, CORBA, LDAP, TMN etc. ）, finally, it proposes an optimized design and implementation of NMS for grid distribute cooperative systems based on JMX and OpenNMS. This implementation can rapidly re-factor current server in a cost-effective manner, and rapidly interface with NMS management, and can precisely identify root causes of failures of distributed collaborative systems.