An Adaptive Configuration Management System Based On Mobile Agent

In recent years, the requirements of large-scale distributed system have increased gradually; system performance and availability are both the aims people pursuit. So, more and more distributed applications adopt redundant servers: on one hand, the system availability could be improved through adding the redundancy of data and computation; on the other hand, user requests are processed on several servers in parallel, so the system performance could be advanced by this means. However, the introduction of redundant servers results in the new challenges to the system configuration management. And it is clear that design a solution to the configuration management of redundant servers. Therefore, it is very important to design a configuration management system for redundant servers so that they could provide more efficient and more rapid service to users.On the basis of analyzing problems of traditional configuration management system, this thesis takes performance, availability and scalability as implement goals, discusses the configuration management for redundant servers, and proposes a mobile agent-based adaptive configuration management system. The features of system are as follow:(1) It adopts hierarchical management architecture to manage redundant servers, so the system scalability is improved and the system overhead is reduced.(2) It adopts RoDO model based on performance index to calculate the redundancy of instances, and this means can make tradeoff between user's performance and availability demands effectively.(3) Form the viewpoint of system performance and availability, ACMS adopts mobile-agent technology to adjust system configuration in running time, improves system's dynamic scalability and resource utilization ratio on condition of guaranteeing the system availability.In the end, a prototype of ACMS is implemented based on CORBA technology, and the overhead and performance of the system is evaluated to prove that ACMS is feasible.