In moving toward an interoperability architecture, the concept of network centric is a step in the right direction-all modules connect to the network, not to each other. And a handful of good network citizenship rules provide a syntactical guide for attachment. From the point of view of the network designer this is sufficient-we have enough to build internetworks for the common good. The continued burgeoning of the Internet constitutes an existence proof. But a common networking base is insufficient to reach a goal of cross-system interoperability-the large information system. Many standardization efforts have attempted to solve this problem, but appear to have lacked the necessary scope. For instance, there have been many efforts aimed at standardizing data elements; these efforts, if followed through, yield some gains, but never seem to quite reach the interoperability goal. If we are to truly erect an interoperability architecture, we need to broaden the scope. This problem of cross-program, cross-service and cross-ally interoperability requires that we agree on the what of modularization, not just the how. This paper is aimed at framing the interoperability architecture problem. On modularization The core of architecture-the way things fit together-is a sense of modularization. This is the part of the problem that is perhaps the least mechanical and requires judgment. Experience, no doubt, helps. Architectural conformity must be traded off against other desired characteristics. The objective is that modules become inherently interoperable so we have components delivered by multiple programs that can be assembled for particular tasks. Prerequisite-network centric.

No abstract