A historical view of HTTP Arena benchmark results, allowing framework maintainers and users to track performance changes over time.
HTTP Arena runs regular benchmarks across dozens of web frameworks and languages, measuring throughput, latency, CPU usage, memory consumption, and more. The results on the site reflect only the most recent run. This project fills the gap by recording every benchmark run into a time series, making it possible to answer questions like:
- Did this release improve throughput?
- When did latency regress, and by how much?
- How does a framework trend compared to its historical baseline?
The project has two components:
Import process — a C# tool that walks the git history of the HTTP Arena repository and extracts benchmark result files from each relevant commit. Results are written as static JSON files under data/<framework>/<test>-<concurrency>.json, deduplicated so that only runs with a meaningful RPS change (or a gap of 7+ days) are recorded.
SPA — a client-side web application that reads the static JSON files and renders interactive time-series graphs. Select a framework, then a test, and explore how any metric has evolved over time.
Because everything is static JSON served from GitHub Pages, no backend infrastructure is required.
Each timeline file has the following structure:
{
"data": [
["2025-01-01T10:22:00Z", { "rps": 43000, "avg_latency_ms": 1.2, "p99_latency_ms": 3.5, ... }],
["2025-01-08T10:22:00Z", { "rps": 43500, "avg_latency_ms": 1.1, "p99_latency_ms": 3.3, ... }]
]
}An index.json at the root of data/ lists all available frameworks and tests for the SPA to enumerate.
dotnet run --project src/Importer/Importer.csproj [<repo-path> [<output-path> [<starting-commit>]]]Defaults point to ../HttpArena as the source repository and ./data as the output directory.
MIT