Setup
Last updated
Last updated
This benchmark is build on the original benchmark for riak core called basho_bench. Additional documentation on basho_bench can be found at basho_bench_documentation.
A benchmark setup for antidote might look as follows:
erlang (for example with kerl)
make
git
r statistics language, to generate graphs
For compiling from source code, clone the repository and use the make command.
git clone https://github.com/AntidoteDB/Benchmarks
cd Benchmarks
make
The compiled basho_bench script is found at the path inside the Benchmark directory:
_build/default/bin/basho_bench
Before you can run a benchmark you have to set up an Antidote cluster, or have access to an existing Antidote cluster. More information on how to deploy Antidote can be found at AntidoteDB-documentation/deployment.
You may use the
make runbench
command, which will use the default results directory, which is the '/tests/' directory inside the repository. If it does not exist it will be created automatically.
If you want to specify your own result directory and/or configuration file, edit the command as follows:
basho_bench --result-dir <results dir> <config file>
The defined results directory contains of benchmarking result folders, that are named as time stamps with the following format:
YYYYMMDD_HHMMSS, for example 20190702_091011
or
current, which is a link to the newest generated result directory
This time stamped directories contain:
the used configuration file: e.g. antidote_pb.config
logs for different log-level
csv files (one summary.csv and maybe several *_latency.csv), which contain the collected data
It is possible to make graphes out of generated csv files.
You can do this by using the following make commands:
make results, which will build the summary graph for the current benchmarking result
make all_results, to build the summary graphs of all benchmarking results contained in the results directory
You can also make these graphs outside of the repository. You only need the '/priv/' folder, as well as the folder containing the results (csv files):
Rscript --vanilla <priv/summary.r> -i <results-directory>
