Sensor Oriented Grid Monitoring Infrastructures for Adaptive Multi-Criteria Resource Management Strategies

In a distributed multi-domain environment, where conditions of resources, services as well as applications change dynamically, we need reliable and scalable management capabilities. The quality of management depends on many factors among which distributed measurement and control primitives are particularly important. By exploiting the extensible monitoring infrastructure provided at the middleware level in a grid meta-scheduling service, in particular integration between GRMS (Grid Resource Management System) and Mercury (Grid Monitoring System), it is possible to perform analysis and then make intelligent use of grid resources. These provide the basis to realize dynamic and adaptive resource management strategies, as well as automatic checkpointing, opportunistic migration, rescheduling and policy-driven management, that has attracted attention of many researchers for the last few years. In this paper we present the current status of our ongoing research in this field together with an example of sensor oriented grid monitoring capabilities facilitating efficient remote control of applications and resources.