A case study in software adaptation

摘要

We attach a feedback-control-loop infrastructure to an existing target system, to continually monitor and dynamically adapt its activities and performance. (This approach could also be applied to "new" systems, as an alternative to "building in" adaptation facilities, but we do not address that here.) Our infrastructure consists of multiple layers, with the objectives of 1. probing, measuring and reporting of activity and state during the execution of the target system among its components and connectors; 2. gauging, analysis and interpretation of the reported events; and 3. whenever necessary, feedback onto the probes and gauges, to focus them (e.g., drill deeper), or onto the running target system, to direct its automatic adjustment and reconfiguration. We report on our successful experience using this approach in the dynamic adaptation of a large-scale commercial application requiring both coarse and fine-grained modifications.