How domgiles/swingbench-public works
Major components
• Front End/Client Tools:
– Swingbench (the graphically rich, main GUI tool)
– Minibench (a simplified, lightweight interface)
– Charbench (the command line version)
• Benchmark Workload Modules:
– Sub-packages for specific benchmarks (e.g., orderentryjdbc, orderentryplsql, json, jsonsocialnetwork, moviestream, saleshistory, sharded, storedprocedures, stresstest, etc.). These represent distinct types of workloads simulated by the engine.
• Schema and Data Setup Wizards:
– Multiple wizards (e.g., oewizard, shwizard, jsonwizard, jsonsocialwizard, moviewizard, tpcdswizard, tpchwizard) located in the “bin” directory (as mentioned in the README). They are responsible for creating the schema and inserting sample data into the database.
• Utility Tools:
– sbutil utility for schema validation, reloading packages, and troubleshooting.
– results2pdf for converting XML test results to a nicely formatted PDF.
– Additional scripts such as parse_results.py for comparing and formatting run results.
Interactions and relationships
• The user (tester or DBA) interacts with one of the front ends (Swingbench, minibench, charbench) to initiate a benchmark run.
• The front end reads the configuration settings (from XML config files as described) and then directs the core benchmark engine to run the specific workload (one of the benchmark modules) against a target database.
• The benchmark modules (each defined in their respective Java packages under src/com/dom/benchmarking/swingbench/benchmarks) execute simulated transactions (e.g., order entry transactions, JSON CRUD operations, social network interactions, movie streaming requests, etc.).
• The data setup wizards (invoked as separate executables) are used before running benchmarks to install the appropriate database schema and populate data.
• Utility tools (like sbutil & results2pdf) act as supporting systems that interface with the database schema or benchmarking results.
Architectural aspects and design principles
• Modular Design: The project is organized into a core “benchmarking engine” with distinct benchmark packages that are modular in nature – each simulating a different workload scenario.
• Front End Abstraction: Having multiple interfaces (GUI and command line) showcases the separation of concerns where the same core benchmark engine is accessible via different interaction modes.
• Configuration-Driven Execution: Workload parameters and database connection details are passed via XML configuration files, indicating a design that allows flexible run-time configuration.
• A note indicating the project’s dependency on Java 17+ and the use of XML for configuration and results output.