One specific problem in wide-area distributed computing environment is effective management of the vast amount of resources that are made available within the grid environment. This paper discusses bandwidth partitioning, fault tolerance, network congestion and scalability issues of grid monitoring and proposes implementation of Dynamic Bandwidth Partitioning algorithm (Q-DBP), on-demand execution of events and agents based monitoring approach respectively to solve these issues. The architecture integrates with the grid middleware layer. This agent infrastructure provides support for mobile agents that are scalable in number and in number of resources.