Scientific visualization serves as a useful technique for rapid analysis of large data sets. The application of visualization techniques to physical simulations executed on highperformance computing hardware offers the opportunity for analysis and design decisions to be performed during simulation execution. This approach, referred to as concurrent visualization, provides some advantages in the present supercomputing environment of shared resources and high-speed networking. We present an implementation of a concurrent visualization system that ties a visualization front-end and user-interface to existing computational fluid dynamics (CFD) simulation code through a light-weight and flexible network transport layer.