As the PC cluster becomes a popular low cost highperformance computing platform, it is hard to manage this system due to the lack of powerful resource management and monitoring tool. This paper presents our effort to mitigate this problem by developing a resource monitoring system. This system also provides a set of API in C, Java and TCL/TK that can be used to develop management applications on top of it. This resource monitoring system has currently been used in our SMILE Beowulf cluster system. We found that this resource management system is a powerful tool for real-time performance monitoring and can be extended later to support dynamic load balancing and task scheduling.