Evolution in Distributed Heterogeneous Systems

Distributed, heterogeneous systems are becoming very common, as globalized organizations integrate applications running on different platforms, possibly written in different languages. Component-interoperability standards such as CORBA are critical enablers of this trend. Certain non-functional requirements for such features as security, quality of service, flexible administration are specially critical to distributed heterogeneous systems. Unfortunately, such requirements are often formulated late, since they depend upon a particular installation, and/or change rapidly with business and political climate. Distributed, heterogeneous systems are particularly difficult to evolve, since the elements are written in different languages, and the operational environment is heterogenous and distributed. Adding “non-functional” features late in the game is specially hard; the required modifications are scattered through the implementations of the different components. Their design and implementation is also obscured by code delocalization, as well as by complexities arising from co-ordination and synchronization considerations. We would like to address this problem with solutions that are animated by practical software engineering goals: type safety of scattered changes, and their interactions; explicit design models, with tracability to code; inter-operability with legacy components and binary COTS components; and opportunistic optimization, leveraging any available optimizations existing in the compilers that are in use.