Towards Solving the Challenge of Minimal Overhead Monitoring - Dataset

This dataset contains the results of a comparison of overhead reduction for the Kieker monitoring framework. The following optimizations were measured using the MooBench framework:

• Using source instrumentation instead of AspectJ,
• Writing into a CircularFifoQueue instead a LinkedBlockingQueue,
• Only storing data of a DurationRecord, instead full metadata storage, and
• Aggregating the stored data.

These have been research with fixed call tree depth 10 (results-kieker-java.tar.xz) and with growing call tree depth (results.tar.xz and results-$size.zip for raw data). Additionally, the plotted runtime growth data depending on the call tree size is contained in the dataset.

The results can be reproduced using ./benchmark.sh (for fixed call tree depth) and ./runExponentialSizes.sh from the folder frameworks/Kieker-java  of the official MooBench repository https://github.com/kieker-monitoring/moobench/ (branch: optimizations-all). The plotting of the runtime growth graph can be repeated using analysis/getExponential.sh.