SQLTraceBench is a tool for benchmarking database performance using trace-based workload generation.
- Trace-Driven: Generates workloads based on real SQL traces.
- Multi-Database Support: Architecture allows plugins for different databases (e.g., MySQL, ClickHouse, StarRocks).
- Statistical Modeling: Uses statistical models to synthesize realistic parameter values.
- Extensible: Easily add new database dialects and workload patterns.
- Visual Reports: Generates comprehensive HTML validation reports with interactive charts.
Check out our Quick Start Guide to get up and running in minutes!
It covers:
- Building the project
- Preparing data
- Running conversion, generation, and benchmarking commands
- Using the automated
examples/quickstart.shscript
SQLTraceBench generates detailed validation reports to compare your benchmark results against a baseline.
Example of an HTML validation report showing QPS deviation and latency distribution.
See Report Interpretation Guide for details on how to read the reports.
- Build:
make build - Test:
make test
See docs/architecture/architecture.md for details on the system design.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
![@google-labs-jules[bot]](https://avatars.githubusercontent.com/in/842251?s=64&v=4){kind=small}
