Abstract: When it comes to creating large distributed systems, many pitfalls await the designers and implementers. Middleware that performs fine in a pilot project suddenly does not scale, bottlenecks appear in unexpected places, and performance drops to a point where the completed system is considered a failure. There are many possible causes for the failure, and they may lie in the middleware as well as the application. This tutorial examines the performance and scalability characteristics of three middleware platforms: Ice, WCF, and RMI. It points out common problems that lie in wait for the unsuspecting distributed system designer and examines design and implementation techniques to avoid them.