Resource Management and Monitoring in AgentTeamwork Grid Computing Middleware

The successful key to resource management in grid computing would include language design for heterogeneous resource description, database design for dynamic resource repository, and tool design for scalable resource monitoring. Particularly focusing on simultaneous use of multiple clusters for grid computing, we are currently addressing these problems by implementing an XML-based resource database and resource-monitoring mobile agents in the AgenfTeamwork grid-computing middleware system. The system describes each cluster in an XML file that even groups cluster nodes based on their resource type. The database returns a collection of clusters or sub cluster groups to a job. The system also deploys a hierarchy of mobile agents to remote nodes for parallel resource monitoring. This paper presents our implementation techniques and preliminary performance of resource management and monitoring in AgentTeamwork.