The increasing need to share resources and information, the decreasing cost of powerful workstations, the widespread use of networks and the maturity of software technologies have increased the demand for more efficient resource management. In this paper we present a decentralized Resource Management System based on the peer-to-peer computing model. Our system coordinates the applications and system resources in an integrated manner; monitors the behavior of the applications transparently and obtains accurate resource projections, schedules the system resources dynamically over multiple machines and reconfigures the objects in response to changing processing and networking conditions. The architecture is based on a multiple feedback loop structure that uses measurements of elapsed time and measurements of processor loads to refine the initial estimates and revise the allocation of the objects on the processors. 18 Pages