lizergsav / camunda-engine-dmn-benchmark

Benchmark for camunda DMN engine

camunda DMN engine - Benchmark

Benchmark for the camunda DMN engine based on JMH.

The benchmark measure how many decision tables can evaluate per second.

How to run it

Go to the project directory and install it with Maven:

$ mvn clean install

After the build is done, you will get the self-contained executable JAR, which holds the benchmark, and all essential JMH infrastructure code. Now, run the benchmark:

$ java -jar target/benchmarks.jar -rf csv -rff camunda-dmn-engine-benchmark-results.csv

This will take a few minutes (~ 5-10min). Then the benchmark is done, you can see the results on the command line and on the generated CSV file camunda-dmn-engine-benchmark-results.csv.

How to customize it

The benchmark class DmnEngineBenchmark have some annotations that configure the benchmark, for example the iterations of measurement. See the JMH docs for details.

Use your own decision table

Put your DMN decision table in the folder src/main/resources/org/camunda/bpm/dmn/engine/benchmark. Configure the benchmark class DmnEngineBenchmark:

• add the key of your decision to the parameter list on the field decisionDefinitionKey
• parse your DMN in the setup method buildEngine() by calling the method parseDecision()
• adjust the variables in the method createVariables() if necessary

How to generate large decision tables

You can use the class DecisionTableGenerator to generate large decision tables. It uses a template and add rules via DMN Model API.