We need benchmarks comparing the behaviour of DN with and without Cumulus4j. The benchmarks should consist of various different data models and queries in order to represent various real-world-scenarios. Of course, they should populate the database with big numbers of records (maybe some ten thousand objects).
The benchmarks should not be part of the regular tests, because they will be very long-running.
The benchmarks should be implemented as integration-tests, i.e. with a jetty server running the DN+Cumulus4j+DB and a client being some junit tests - a separate key server is optional - but some benchmarks should definitely use it. Or maybe we even compare every scenario with and without key server?! Have to think about an easy way to accomplish this.