The Design And Implementation Of JMX-based Remote Job Management System

Nowadays the increasing network applications generate new requierments for high performance network servers. Compared with traditional high performance computer, cluster system is gradually becoming prevailing network server because of its excellent value for money feature and high extensibility. A cluster server is composed of many computing nodes, and many types of jobs can be run on cluster. In addition, the development of Grid Computing further extends the application of traditional cluster system. It is, therefore, very important to manage the whole system and to schedule jobs in grid environment in a consistent way. The goal of this paper is to do indepth research on how to design and implement secure, extensible, and dynamicly controllable remote job management system:Firstly, due to the insufficient extensibility and manageability in traditional system, this paper emphatically presented a RJMS Management Framework ("RJMS MF"). RJMS MF takes JMX as its microkernel and supports dynamic management of grid and cluster. It can enhance the extensibility and manageability in traditional system by providing the consistent management method of system components, resources, services, management interfaces and persistent services.Secondly, aiming at the problem of lacking security in traditional system, we presented and implemented RJMS Security Framework ("RJMS SF'). RJMS SF contains Enterprise Authority Mapping Model ("EAMM). EAMM can easily map users in local system (such as OA system, etc) to virtual users in grid environment. And at the same time, RJMS SF can ensure the security of the whole data channel in RJMS system.Thirdly, we presented a two-layer scheduling framework which is suitable for grid-cluster environment. This scheduling framework supports schedule policy based on complex formula and dynamic policy choosing mechanism. Meanwhile, dynamic management and policy hot-deploy is also supported in this framework.The design of RJMS Framework presented in this paper has been applied in the Remote Job Management System of Cluster Server Function Software with excellent performance in system functions, security and management. The research provides high application value for designing and implementing job management system in grid computing environment with security, management and scheduling supports.