December 1, 2003 - December 1, 2005
During the past two attempts to demonstrate our highly distributed majordemo system we experienced many low level problems that prevented the demo from going smoothly. These ranged from machines being off the network, to agents not being in the right state, to speakers being unplugged. It is well known that keeping a complex distributed system up and running is a difficult task, made even all the more difficult if many of the components are partially or completely autonomous. This project will begin by investigating the techniques and approaches that have been used to develop robust and "self-repairing" autonomic distributed systems and then select a set of them to integrate into our our systems (e.g., majordemo, cobra, etc.). When possible, these techniques would be added at a very general level, e.g., to the infrastructure level or as new agents or web services. The goal would be to actually improve the robustness and performance of our working systems. For more information, contact Tim Finin, finin@Umbc.edu.