Benchmarking Change Detection Exactness and Overhead of Instrumentation and Sampling - Dataset
收藏资源简介:
This contains the code and the experimental results for the paper "Benchmarking Change Detection Exactness and Overhead of Instrumentation and Sampling". Especially, it contains 3 things Code code.tar: The code of sjsw, peass, precision-experiments, and precision-experiments-rca that are needed to execute the experiments (although the GitHub repos are preferable, if they are available) To run the experiments, build all projects as described in their READMEs and run for percentage in 1 2 3 4 5; do ./run_instrumentation_regular.sh $percentage UNTIL_SOURCE_CHANGE; done for percentage in 1 2 3 4 5; do ./run_sampling.sh $percentage 1; done for percentage in 1 2 3 4 5; do ./run_sampling.sh $percentage 10; done for percentage in 1 2 3 4 5; do ./run_sampling.sh $percentage 20; done ./run_measure.sh 1 The experiments can be distributed, as long as the resulting _peass folders are copied to the same analysis folder For analysis, use the scripts in precision-experiments-rca/scripts/analysis_instrumentationSampling Measurement Data instrumentation-usc.tar: Measurement results of instrumented executions of the demo project, where only the methods that are calling the method with changes source code are instrumented sampling-interval1.tar.zstd: Measurement results of the executions using sampling with 1 ms interval sampling-interval10.tar.zstd: Measurement results of the executions using sampling with 10 ms interval sampling-interval20.tar.zstd: Measurement results of the executions using sampling with 100 ms interval Analysis results results.tar: Heatmaps and analysis results based on the data overhead.pdf depicting the overhead of run without instrumentation, with instrumentation and with sampling
本数据集包含论文《基准测试程序插装与采样的变更检测精确性及运行开销》("Benchmarking Change Detection Exactness and Overhead of Instrumentation and Sampling")的配套代码与实验结果,具体包含以下三类内容: 一、代码 code.tar:包含运行实验所需的sjsw、peass、precision-experiments及precision-experiments-rca代码(若对应GitHub仓库可正常访问,优先使用仓库源码)。 如需运行实验,请按照各项目README文件中的说明完成构建,并执行以下命令: for percentage in 1 2 3 4 5; do ./run_instrumentation_regular.sh $percentage UNTIL_SOURCE_CHANGE; done for percentage in 1 2 3 4 5; do ./run_sampling.sh $percentage 1; done for percentage in 1 2 3 4 5; do ./run_sampling.sh $percentage 10; done for percentage in 1 2 3 4 5; do ./run_sampling.sh $percentage 20; done ./run_measure.sh 1 实验可分布式执行,只需将生成的_peass文件夹复制至同一分析目录即可。分析环节请使用precision-experiments-rca/scripts/analysis_instrumentationSampling目录下的脚本。 二、测量数据 instrumentation-usc.tar:演示项目的程序插装(Instrumentation)执行测量结果,仅对调用了含变更源码的方法进行插装。 sampling-interval1.tar.zstd:采用1ms采样(Sampling)间隔的执行测量结果。 sampling-interval10.tar.zstd:采用10ms采样间隔的执行测量结果。 sampling-interval20.tar.zstd:采用100ms采样间隔的执行测量结果。 三、分析结果 results.tar:基于上述数据生成的热图与分析结果。 overhead.pdf:展示无插装、仅插装及采用采样三种场景下的运行开销。



