Dynamic Evolution Software Model In Service-oriented Systems

In order to adapt to the opening environment of Internet and the variable requirements of users, software systems should be able to adjust themselves. The adjustment, or evolution, can be static or dynamic. Compared with static evolution, dynamic evolution has many advantages, although it is technologically more complex. Dynamic evolution technology is a research focus in academy and industry.Software systems are always constructed by a number of refinement steps, starting from user requirement. Refinement process forms a pyramid like structure with every layer representing software architecture, which is business logic composed by components and connectors. Different software architectures have different refinement levels, and refinement relationship exists between them.Traditional research of dynamic evolution focuses on evolution of software architecture on single layer, and especially the bottom-layer software architecture. This thesis attempts to handle this issue from the process of software construction, and to research on the dynamic evolution property of multi-layer software architectures. We apply context sensitive grammar to the background of component computing and software refinement, and propose a specific definition of context in that application. We propose a dynamic evolution software model in service-oriented systems-DESM. DESM can gracefully handle user requirement change, rule change and network condition change. In DESM, users specify meta-specification and constraint to filter context-sensitive rules. We propose an algorithm of matching constraint and rule context. Rule change is configured by user, and triggered by system. We discuss the constraint and format of rule change. For network condition change, we propose a mechanism called component agent to reduce cost and to make change transparent to upper-level software architecture.DESM is one part of Component-Based Software Integrated Development Environment (CBSIDE). It can dynamically combine services, display the runtime status of service combination, and allow user to reconfigure. For that purpose. we designed three modules: Requirement Refinement Module, Execute Engine Module. and Component Availability Monitor.The theoretical and implementation work of this thesis is a base for furtner research on software dynamic evolution technology and implementation of CBSIDE.