This paper focuses on an implementation of process farms under the parallel virtual machine environment. PVM uses a cluster of workstations to form a heterogenous network based MIMD computing environment. The problems associated with such an environment to form a parallel application are that, typically point to point communication constructs are provided to distribute data in a data parallel approach. Most parallel applications need some form of load balancing technique to equitably distribute work packets and keep all processors busy. This is usually a static off line process in which the application is evaluated for a particular problem size and a log is examined to determine the placement of tasks on the nodes by examining the computation to communication time ratios for the tasks. The application has to explicitly distribute the data so as to have equal load on all processors.