This is the first version of the node benchmark.
Note: Initial transactions itself does not affect the CPS (confirmations per second) calculation—only the actual benchmark transactions and their subsequent confirmations are measured.
Currently, the benchmark focuses on a single node. This choice reduces database load because all nodes share the same database. In earlier attempts, I tried running one database per node, but the database quickly became a bottleneck. Using a single database made the benchmark run faster overall, though it doesn’t fully reflect real-world behavior. To keep things simple for now, I decided to focus on a single-node setup rather than testing multiple nodes on the same machine.
./gradlew run
Requirements: Docker or Podman