Dynamic Reconfiguration for Middleware-Based Applications

Distributed systems with high availability requirements have to allow reconfiguration of the system without being taken off-line. Examples of reconfigurations are the replacement of a component with a newer version, or the migration of a component to another node. A key issue for reconfiguration is maintaining the correctness of the system, which can be very complex due to the number of components, unclear relations between components, heterogeneity in operating systems and programming languages, and physical distribution of components. In this paper, we describe a new approach for dynamic reconfiguration of middleware-based applications that is more transparent for the application developer than existing approaches. We compare our approach with other approaches, and describe a prototype that implements our approach for CORBA-based applications. Index Terms – Dynamic reconfiguration, distributed systems, middleware, CORBA 1 Submitted to IEEE Transactions on Parallel and Distributed Systems, Special Issue on Middleware (Fall 2003).