kTBS Bench

This project aims at exploring performance bottlenecks of the kTBS.

It is made of:

• custom libraries for benchmarking (see also sutils)

• scripts to test some performance scenari

• results graph and their associated notes and explanations

Results

Reports on our findings are :

• [Report on query benchmarks] report-query

• [Report on insert benchmarks] report-insert

More information is available via the online documentation.

How to setup this project

1. Clone this repository using git.

2. Make it a python virtual environment by cding to the project directory and virtualenv ..

3. Activate the virtual environment: source bin/activate.

4. Install the necessary python packages: pip install -r requirements.txt.

   If you don't want to run the examples, you should uncomment the lines corresponding to the heavy libraries pandas, matplotlib and numpy in requirements.txt.

Running benchmarks

You can run the benchmark example like so: python bench_examples/bench_queries.py.

You can also define your own benchmarks. See bench_examples/bench_queries.py for a template. You might want to read the [documentation of ktbs_bench_manager] kbm-doc.

Furthermore, you can run a batch of benchmarks by using the bench.py script. Use bin/bench.py --help for more information.

Documentation

The documentation for this project is available on readthedocs. The reports on our findings are there too.

License

MIT, see LICENSE file.