Grid Computing Research Laboratory State University of New York (SUNY) Binghamton Department of Computer Science Michael J. Lewis (mlewis@binghamton.edu) Co-Director, Assistant Professor of Computer Science Madhusudhan Govindaraju (mgovinda@.cs.binghamton.edu) Co-Director, Assistant Professor of Computer Science

Dynamically Configurable Distributed Objects

The dynamically configurable distributed object (DCDO) model helps enable object evolution and facilitate the development of distributed objects from multiple independent implementation components. Using DCDOs, programmers can evolve existing active objects to accept new member functions, to change the interface and behavior of their member functions, and to remove member functions from their public or private interface. Programmers can make these changes on the fly, without deactivating any part of the system, without replacing binary executables, without interrupting the clients of evolving objects, and without having to know what the changes will be at the time the objects are initially compiled and run. The model supports evolution management strategies that define when and how object types evolve from one version to the next, and determine when a type change is propagated to existing instances.

DCDOs represent a useful way to evolve active distributed objects in a wide area distributed object computing system, and to enable component-based object composition via the mechanisms of the host distributed system itself. The DCDO model describes the functionality and roles of the object types necessary for an implementation of dynamic configurability. The model is simple, defining the interfaces to just three different object types, it is independent of any particular programming language, operating system, or architecture, and it is flexible enough to support a range of different evolution management policies and strategies.

The complete DCDO model has been fully implemented within the current version of the Legion wide-area distributed object computing system. A performance study of this implementation measures the run-time overhead associated with the DCDO mechanism, and compares the performance of evolving DCDO objects with that of evolving normal Legion objects. This study demonstrates that dynamic configurability can be implemented efficiently, with a small amount of added overhead for method function invocation and object creation, and with a marked improvement in the cost of evolution.

For more information on the DCDO model please read the following paper: