Dynamic Software Product Line for Improving Energy Efficiency of Cloud Applications

The dataset is structured around two feature model types under study: SPLOT Repository and TeaStore. For each feature model, the data is organized into three folders:- FeatureModels/ contains the XML (for SPLOT) and JSON files that define the configuration space explored during the experiments.- ExecutionMetrics/ contains one subfolder per experimental benchmark and algorithm (named <timestamp>), each holding a summary.csv with run-level resource metrics (CPU time, memory, exit codes) alongside individual run_<N>/ directories storing raw execution artifacts such as job scripts, standard error logs, and timing files.- EnergyConsumptionData/ is organized hierarchically by feature model, algorithm, and run index. For each combination, there's one energy_<N>.json file per run, raw watt-metre measurements retrieved from the Grid5000 infrastructure API; alongside a summary_<N>.csv capturing the derived energy metrics for that run. A global_summary.csv at the root of this folder aggregates all runs across feature models and algorithms.