Run the Benchmark

Download the benchmark kit from your unlock page and transfer it to your Linux host.

scp dataforge-benchmark-kit.tar.gz user@host:~/
ssh user@host
tar -xzf dataforge-benchmark-kit.tar.gz
cd dataforge-benchmark-kit/

The setup script fingerprints your hardware and database environment. It saves a JSON snapshot used later to contextualize results.

./scripts/bench_setup.sh \
  --conn "postgres://user:pass@localhost/benchdb" \
  --device nvme0n1 \
  --netif eth0

Replace --device with your block device (e.g. sda, dm-3, nvme0n1). Use lsblk to identify it.

The suite runs DataForge at increasing concurrency levels (1, 2, 5, 10, 20, 50). Each level ingests the full dataset and records metrics. Allow 15–30 minutes for the full run.

./scripts/bench_suite.sh \
  --url http://localhost:8080 \
  --key YOUR_API_KEY \
  --conn "postgres://user:pass@localhost/benchdb" \
  --dir /path/to/csv/files \
  --prefix citation-map \
  --device nvme0n1

bench-report reads the JSON output from the suite, runs live Stage 3 database validation, and generates a self-contained HTML report.

./bench-report \
  --results-dir bench_results_YYYYMMDD_HHMMSS \
  --conn "postgres://user:pass@localhost/benchdb" \
  --prefix citation-map \
  --count 50 \
  --expected-rows 1516282 \
  --out my-bench-report.html \
  --tag "My Infrastructure — 2026-03-26"

The generated HTML file is self-contained. Open it in any browser. It includes your environment fingerprint, concurrency ladder results, hardware pressure metrics, and Stage 3 data validation.

open my-bench-report.html