Benchmarking Utility

Contains two scripts: build.py and benchmark.py

Installation

This requires cnx-archive be installed.

It also requires these third-party dependencies:

pip install docopt gitpython

Usage

Build a set of cached queries using one of the following commands:

python build.py "dbname=cnxarchive-testing user=cnxarchive" \
--repo=../cnx-archive/ "carpentry"

Or:

python build.py "dbname=cnxarchive-testing user=cnxarchive" \
--repo=../cnx-archive/ \
--from-data-set=./cache/ff781b8ab034866eb0b36adc10e76abc75144601/data-set.json

After the data-set has been build into a cache of SQL statements you can run the benchmark tool over the data-set to observe the time it takes to get back results from the raw SQL statements.

python benchmark.py "dbname=pumazi user=pumazi password=pumazi" --data-set=cache/ff781b8ab034866eb0b36adc10e76abc75144601/data-set.json

This will print the times out in the console as well as produce a CSV file of the results for later analysis.