With the spread of networks of workstations and the wide-range of application communication and computation requirements, we may ask of each system: is the potential application bottleneck in the processor or the network? We propose an experimentally based model for visualizing and measuring the capacity of a system's processor and network. We apply our model to three typical networks of workstations: home, academic and research. Based on our analysis, we find that research clusters of high-performance workstations may still have computation bottlenecks when they are connected with the highest bandwidth networks. Academic clusters of Sun workstations, on the other hand, have the network as a potential application bottleneck. Home networks of personal computers have the network as a severe potential application bottleneck. Our model may direct work in application development and network of workstation design by exposing potential bottlenecks.