experiments

TraceMOP experiments

Projects and data

You can find our 105 projects here, and you can find our experiments data in this directory.

Generate CSV

To generate the plots we have in the paper, run

pip3 install matplotlib numpy
python3 make-project-data.py data/data-actual.csv .

Usage

Prerequisites:

• A x86-64 architecture machine
• Ubuntu 20.04
• Python 3.9 or above
• Docker

Run experiments

# Inside this directory, outside Docker container,
# run the below command to run TraceMOP without trace tracking and JavaMOP (could take multiple days)
bash run_javamop_in_docker.sh tracemop-vs-javamop.txt tracemop-vs-javamop

# Inside this directory, outside Docker container,
# run the below command to run TraceMOP with trace tracking and our previous prototype (could take multiple days)
bash run_in_docker.sh tracemop-vs-prototype.txt tracemop-vs-prototype

# After running both commands above, run the below in the same directory
bash collect-results.sh tracemop-vs-javamop tracemop-vs-prototype

Outputs

After running the above experiments, the current directory will contain 4 csv files with the following header:

tracemop-vs-javamop-time.csv: project name, time to download project, time to run TraceMOP without trace tracking, time to test project, and time to run JavaMOP (in ms)

tracemop-vs-prototype-time.csv: project name, time to run TraceMOP with trace tracking, time to run our previous prototype (in ms)

tracemop-vs-prototype-memory.csv: project name, memory used by TraceMOP with trace tracking, and memory used by our previous prototype (in KB)

tracemop-vs-prototype-disk.csv: project name, disk used by TraceMOP with trace trackin, and disk used by our previous prototype (in byte)