This paper discusses issues related to a network-based catalog part selection system. Some of the key issues are: the optimal search problem, the burden imposed on the server of a network-based catalog or the time spent in data retrieval, identical parts in different vendor catalogs, and the excessive amount of data in a database of component information. The paper addresses the problem of network “burden,” the problem of identical parts, and the excessive data problem. The authors employ a “Configuration-decomposition” strategy in the approach to the excessive data problem. As an example of applying this approach, the implementation of the authors’ experimental system for part selection is discussed.