Survey on Concern Separation in Service Integration

Abstract

Ever-changing business processes in large software systems, integration of heterogeneous data sources as well as the desire for legacy service integration drive software design towards reusable, platform-independent, web-accessible microservices. Such independently deployable services provide an interface for retrieval and data manipulation in machine-readable formats. While this approach brings many advantages from the perspective of service integration aiming to separate data manipulation from business processing, the standard approaches provide only limited structural semantics and constraints provided through the interface. This leads to considerable information restatement and repeated decisions in integrating components, which considerably impacts development and maintenance efforts. Integration component operability becomes highly sensitive to interaction with underlying services, which are possibly composed of other services. The sensitivity is especially apparent in the structural semantics of produced and consumed information that must correlate on both sides of the interaction. This paper surveys service integration from the perspective of separation of concerns. In order to reduce the coupling and information restatement on the integration component side, it suggests introducing multiple communication channels with additional information that apply in the service interaction, extending the integration component’s ability to derive service expected information structural semantics, constraints or business rules. Finally, we consider the impact of this new approach from the development and maintenance perspectives.