Online auctions are now an immensely popular component of the electronic marketplace. However, there are many fraudulent buying/selling behaviours that can occur during an auction (e.g., shill bidding, bid shielding, etc.). While researchers are proposing methods for combating such fraud, it is extremely difficult to test how effective these countermeasures are. This is primarily due to it being unethical to engage in fraudulent behaviour just for the purpose of testing countermeasures. Furthermore, there is limited commercial auction data available due to the sensitivities of an online auctioneer being willing to admit that fraud has, or is occurring. In order to test fraud countermeasures in a controlled environment, we have created our own online auction server for conducting auction-related research. This paper presents our experiences with designing and implementing our own online auction system which we call uAuction. At present, there is limited useful literature on auction system design. We present an analysis and design of the auction system by employing Unified Modeling Language (UML) to show the architectural model, subsystems, use cases, activity workflows, class diagram, user interfaces, and system sequence diagrams. Our auction model is grounded in object-oriented techniques and is open source so that other researchers can expand upon our approach.