ChunkLoris

research

A cross-ecosystem measurement study and replication harness for per-chunk HTTP request amplification in production servers

period:2026-present
tech:
HTTPApplication SecurityPerformance MeasurementDockerReproducible Research

ChunkLoris measures the CPU cost created when valid request bodies are delivered as many one-byte chunks or frames. The project compares production HTTP/1.1 servers across seven language ecosystems and extends the measurement shape to HTTP/2, HTTP/3, and WebSockets.

The repository includes the working paper, source-cited measurement matrix, standardized probe clients, per-server Docker contexts, raw JSON results, schemas, figure-generation scripts, and an arXiv-ready source package. It distinguishes measured behavior from vulnerability claims and documents buffering and scheduling mitigations across server and proxy designs.