Run benchmarks

Use benchviz to run k6 and generate reports from the benchmark stack.

For the public result narrative, see Benchmark results. This guide is about producing and reviewing benchmark artifacts.

Install dependencies

npm --prefix benchmark install

Run the default sweep

npm --prefix benchmark run benchviz -- run \
  --stack khone-benchmark \
  --region us-east-1 \
  --duration 3m \
  --stage-targets 50,100,150 \
  --max-delay-ms 0

Run a smoke test

npm --prefix benchmark run benchviz -- run \
  --stack khone-benchmark \
  --region us-east-1 \
  --run-name smoke \
  --duration 5s \
  --stage-targets 1 \
  --executor ramping-vus \
  --max-delay-ms 0

Render from an existing CSV

npm --prefix benchmark run benchviz -- run \
  --skip-test \
  --csv-path benchmark-results/some-run/k6.csv \
  --run-name rerender \
  --themes light-transparent,dark-transparent

Generate a shareable report bundle

Use a stable, descriptive --public-run-name for report bundles you want to keep or share. Avoid generated timestamp directory names when the result should be easy to compare later.

npm --prefix benchmark run benchviz -- run \
  --skip-test \
  --run-dir benchmark-results/some-run \
  --public-run-name high-256mb-min4-50-100-150-round2 \
  --public-output-dir benchmark-results-public \
  --themes light-transparent,dark-transparent

The report bundle contains:

report.md
summary.csv
charts/light/latency-distribution.png with light colors and a transparent background
charts/dark/latency-distribution.png with dark colors and a transparent background

The raw run directory still contains additional files such as run.json, k6.csv, benchviz.log, and data/metrics.json.