Understanding the Architectural Characteristics behind Middleware Choices

Integrating heterogeneous components to form a distributed system can manifest difficult interoperability problems among the components. Contributing to this problem is a lack of understanding of the underlying software architecture of the components, and, possibly, the distributed architecture in which they participate. Middleware to resolve these problems may be difficult to choose and implement, often resulting in integration solutions that are not complete or evolvable. This paper discusses software architectural characteristics that underlie choices for integration solutions to interoperability problems in distributed component architectures. The 4+1 view model of architecture is used to represent the characteristics at their appropriate level of abstraction. We use two distinct case studies to illustrate how comparisons among characteristic values of participating components in the integrated system could have predicted interoperability problems.