Managing evolution in distributed systems

Distributed computer systems offer the potential for growth and change. Although the underlying support mechanisms for change (software creation, binding, deletion) are well understood, there has been no consensus on how evolutionary change can be managed and controlled. This paper suggests that, since distributed systems are specified and constructed in terms of their software structure (configuration), it is appropriate that changes should be specified in terms of structure and managed at the configuration level. We present a model for managing evolution which separates structural concerns at the configuration level from those application concerns at the component level. This permits the formulation of general structural rules for change without the need to consider application state, and the specification of application actions without knowledge of the actual structural changes which may be introduced. To illustrate the approach and discuss pragmatic aspects, the model is applied to an example problem which has been implemented and tested in the Conic environment for distributed systems.