Welcome to BenchBox Discussions

[joeharris76]

Thanks for joining the BenchBox community! This is the place to ask questions, share experiences, and discuss database benchmarking with fellow users and contributors.

What is BenchBox?

BenchBox is a benchmarking toolbox that makes it simple to benchmark analytical (OLAP) databases. It provides industry-standard benchmarks (TPC-H, TPC-DS, ClickBench, and more) in a single Python package with embedded data generation and cross-platform support.

Key features:

• 18 embedded benchmarks including TPC-H, TPC-DS, SSB, ClickBench, and more
• Cross-database support: DuckDB, Snowflake, Databricks, BigQuery, ClickHouse, Polars, and more
• DataFrame mode benchmarking with Polars, Pandas, and other libraries
• Automatic SQL dialect translation between platforms
• Simple CLI: benchbox run --platform duckdb --benchmark tpch
• MCP server for AI agent integration (Claude Code, etc.)

Getting Started

# Install
uv add benchbox
# or: pip install benchbox

# Run your first benchmark
benchbox run --platform duckdb --benchmark tpch --scale 0.01

See the documentation and Getting Started guide for more details.

MCP Server for AI Agents

BenchBox includes an MCP (Model Context Protocol) server that enables AI assistants like Claude Code to run benchmarks directly through natural language. Install with uv add benchbox --extra mcp and run:

benchbox-mcp

The MCP server provides tools for:

• Running benchmarks and dry runs
• Listing platforms and benchmarks
• Comparing results across runs
• Analyzing performance patterns
• System profiling for capacity planning

See the MCP Integration Guide for setup instructions.

Discussion Categories

• Announcements - Project updates and releases
• Q&A - Get help with installation, usage, or troubleshooting
• Ideas - Suggest new features, benchmarks, or platforms
• Show and Tell - Share your benchmarking results, workflows, or integrations
• General - Everything else

Community Guidelines

• Search first - Check existing discussions before starting a new one
• Be specific - Include platform, benchmark, scale factor, and error messages when asking for help
• Be constructive - Focus on problems and solutions
• Be patient - BenchBox is alpha software maintained as a personal project

Alpha Status

BenchBox is currently in alpha. APIs may change, and some features are incomplete. Your feedback and bug reports help shape the project's direction.

Links

• GitHub Repository
• Documentation
• Bug Reports & Feature Requests
• PyPI Package