Shadow

Spec sheet

Purpose

Shadow is a scientific experimentation tool for research, development, testing, and evaluation of real networked applications on simulated networks.Read 8 Oct 2026 · shadow.github.io

Privacy

Shadow says simulations are private and segregated from the Internet.Read 8 Oct 2026 · shadow.github.io

Support

The project directs users to GitHub Discussions for community support and GitHub Issues for bug reports.Read 8 Oct 2026 · github.com

Configurable networks

A configurable network graph models characteristics such as path latency and packet loss.Read 8 Oct 2026 · shadow.github.io

Dependencies

Building Shadow requires Linux development dependencies including GCC, Python, GLib, CMake, and Rustup.Read 8 Oct 2026 · shadow.github.io

Deterministic experiments

Experiments can be scientifically controlled and deterministically replicated to help reproduce bugs and reduce confounding factors.Read 8 Oct 2026 · shadow.github.io

Example workflow

The basic example simulates an HTTP server and three clients making requests across virtual hosts.Read 8 Oct 2026 · shadow.github.io

Execution model

Shadow runs real, unmodified application binaries as native Linux processes in a discrete-event simulation.Read 8 Oct 2026 · shadow.github.io

Network controls

Shadow models routing characteristics such as path latency and packet loss using a configurable network graph.Read 8 Oct 2026 · shadow.github.io

Network model

Shadow connects managed applications through an internal simulated network with protocols such as TCP and UDP.Read 8 Oct 2026 · shadow.github.io

Network simulation

It connects managed processes through a private virtual network and simulates protocols including TCP and UDP.Read 8 Oct 2026 · shadow.github.io

Open source

The project describes Shadow as 100% open-source software written primarily in Rust and C.Read 8 Oct 2026 · shadow.github.io

Real application execution

It runs unmodified application binaries as native Linux processes inside a discrete-event simulation.Read 8 Oct 2026 · shadow.github.io

Reproducibility

Shadow experiments are deterministic, so bugs can be reproduced by rerunning a simulation.Read 8 Oct 2026 · shadow.github.io

Scale

Shadow says it can simulate distributed systems with thousands of network-connected processes on a Linux laptop, desktop, or server.Read 8 Oct 2026 · shadow.github.io

Security caveat

Shadow's security policy warns that it is not designed to isolate potentially malicious programs, which may access host files or issue native system calls.Read 8 Oct 2026 · github.com

Source code

Shadow is 100% open-source software written primarily in Rust and C.Read 8 Oct 2026 · shadow.github.io

Supported use case

The documentation says large-scale Tor network simulations are a particularly well-supported use case.Read 8 Oct 2026 · shadow.github.io

System call limits

Shadow implements over 150 system call API functions but does not yet fully support all API features, so applications using more complex features may not work correctly.Read 8 Oct 2026 · shadow.github.io

System call support

Shadow implements over 150 system call API functions, but applications using unsupported or complex features may not work correctly.Read 8 Oct 2026 · shadow.github.io

Testing and debugging

The project describes using large, diverse test networks to exercise corner cases and improve test coverage.Read 8 Oct 2026 · shadow.github.io

Tor use case

The documentation says large-scale Tor network simulations are a particularly well-supported use case.Read 8 Oct 2026 · shadow.github.io

Where Specifiction places it

• Network Testing Software#29 of 336.8
• Network Simulation Software#31 of 316.7