ninedata/dbbench repository overview

⁠NineData DBBench

Ready-to-Use TPC-C Database Benchmark Tool

DBBench is a professional database performance testing tool that fully implements the TPC-C benchmark specification. It supports 12 major databases and provides a real-time web monitoring dashboard.

⁠Image Highlights

• Multi-Database Support: MySQL, PostgreSQL, Oracle, SQL Server, DB2, TiDB, OceanBase, Dameng, SQLite, YashanDB, Sybase, SAP HANA
• Real-Time Monitoring: Web UI with live tpmC, latency, CPU, memory, network metrics and time range selector (All / 1m / 10m / 1h / 6h / 1day / Custom)
• Full TPC-C: All 5 transaction types (New-Order, Payment, Order-Status, Delivery, Stock-Level)
• CSV Fast Load: Database-native bulk import (MySQL LOAD DATA, PostgreSQL COPY, etc.) for faster data loading
• Flexible Configuration: Environment variable support for various testing scenarios

⁠Quick Start

docker run -d -p 1929:1929 --name dbbench ninedata/dbbench:latest

Open http://localhost:1929⁠ in your browser to start testing.

⁠Main Dashboard

⁠Configuration Panel

⁠Supported Databases

⁠Usage Workflow

1. Start Container: Run the docker run command
2. Access Web UI: Open http://localhost:1929⁠
3. Configure Connection: Set database connection in the config panel
4. Load Data: Click "Load Data" to generate test data
5. Run Benchmark: Click "Start" to begin the benchmark
6. View Results: Monitor TPS, latency, and system resources in real-time, use time range selector to view different periods
7. Export Report: Click "Report" after test completes to view/export the benchmark report

⁠Data Scale Reference

Each warehouse contains approximately:

• Customer: 30,000 rows
• Order: 30,000 rows
• Order-Line: ~300,000 rows
• Stock: 100,000 rows

Recommended Configurations:

• Quick Test: 1-2 warehouses, 5-10 terminals
• Standard Test: 10 warehouses, 50 terminals
• Stress Test: 100+ warehouses, 200+ terminals

⁠Port Reference

⁠REST API

⁠Metrics Collected

⁠Transaction Metrics

• Throughput (tpmC / TPM)
• Total transactions count
• Success / Rollback / Error counts and rates
• Latency (average, min, max)
• Per-transaction type breakdown

⁠Database Host Metrics (via SSH)

• CPU usage (%)
• Memory usage (%)
• Disk I/O (read/write bytes/sec)
• Network I/O (recv/sent bytes/sec)

⁠Database Metrics

• Active connections
• Buffer pool / Cache hit ratio
• Row lock waits
• Slow queries count

⁠Client Metrics (Local Machine)

• CPU usage (%)
• Memory usage (%)
• Load average (1m, 5m, 15m)
• Network I/O (bytes/sec)
• Disk I/O (bytes/sec)

⁠CLI Mode

In addition to the Web UI, CLI mode is also supported using configuration files:

# Enter container
docker exec -it dbbench bash

# Load data
java -jar /app/dbbench.jar -f /app/profiles/config.properties load

# Run benchmark
java -jar /app/dbbench.jar -f /app/profiles/config.properties run

# Combine actions: clean → load → run
java -jar /app/dbbench.jar -f /app/profiles/config.properties clean load run

# Override specific settings
java -jar /app/dbbench.jar -f /app/profiles/config.properties -w 20 -c 100 -d 300 run

⁠Source Code & Documentation

• GitHub: https://github.com/ninedata-cloud/dbbench⁠
• Issues: https://github.com/ninedata-cloud/dbbench/issues⁠

⁠License

Apache License 2.0