Academic Commons

Reports

Combining Mobile Agents and Process-based Coordination to Achieve Software Adaptation

Valetto, Giuseppe; Kaiser, Gail E.

We have developed a model and a platform for end-to-end run-time monitoring, behavior and performance analysis, and consequent dynamic adaptation of distributed applications. This paper concentrates on how we coordinate and actuate the potentially multi-part adaptation, operating externally to the target systems, that is, without requiring any a priori built-in adaptation facilities on the part of said target systems. The actual changes are performed on the fly onto the target by communities of mobile software agents, coordinated by a decentralized process engine. These changes can be coarse-grained, such as replacing entire components or rearranging the connections among components, or fine-grained, such as changing the operational parameters, internal state and functioning logic of individual components. We discuss our successful experience using our approach in dynamic adaptation of a large-scale commercial application, which requires both coarse and fine grained modifications.

Subjects

Files

More About This Work

Academic Units
Computer Science
Publisher
Department of Computer Science, Columbia University
Series
Columbia University Computer Science Technical Reports, CUCS-007-02
Published Here
April 21, 2011