P4TG: 1 Tb/s Traffic Generation for Ethernet/IP Networks

• Overview
• Installation & Start Instructions
  • Data plane
  • Control plane
  • Configuration GUI
• Changelog
• Documentation
• Preview

This repository contains the source code for a P4 based 1 Tb/s traffic generator based on the Intel Tofino(TM) ASIC, called P4TG (Paper). The paper version corresponds to v.1.0.0.

In generation mode, P4TG is capable of generating traffic up to 1 Tb/s split across 10x 100 Gb/s ports. Thereby it measures rates directly in the data plane. Generated traffic may be fed back from the output to the input ports, possibly through other equipment, to record packet loss, packet reordering, IATs and sampled RTTs. In analysis mode, P4TG measures rates on the input ports, measures IATs, and forwards traffic through its output ports.

P4TG (v2.0.0) supports VLAN (802.1Q) and QinQ (802.1ad) encapsulation. P4TG (v2.1.0) further supports MPLS.

P4TG consist of:

• a P4 program for the Intel Tofino(TM)
• Rust control plane
• React configuration GUI

Go to P4-Implementation and compile p4tg via make compile. This compiles the program and copies the resulting configs to the target directory.

Afterwards, start p4tg via make start.

This requires a fully setup SDE with set $SDE and $SDE_INSTALL environment variables.

Tested on:

• SDE 9.9.0 (up to v2.0.0)
• SDE 9.13.0

The controller is written in rust and can be started via docker-compose up. The initial build may take a few minutes.

The controller then starts a REST-API server at port P4TG_PORT and endpoint /api (see docker-compose.yml) that is used to communicate with the configuration GUI. It also serves the configuration gui at port P4TG_PORT and endpoint /.

Set SAMPLE=1 in docker-compose.yml to activate IAT sampling mode instead of data plane measurement. Data plane measurement mode (SAMPLE=0) is more accurate and the default

The configuration GUI is based on react & nodejs. It is automatically served by the controller at http://ip-of-tofino-controller:P4TG_PORT.

If you want to adapt the configuration GUI, re-build the configuration GUI via npm run build within the Configuration GUI folder and copy the build folder to Controller/gui_build. Afterward, re-build the controller via docker-compose build within the Controller folder.

The configuration GUI can also be run independently of the controller. Adapt the API_URL in the config.ts if the controller and configuration GUI run on different machines.

To run the configuration GUI via docker-compose run docker-compose up. After the build has finished, the configuration GUI is reachable at http://127.0.0.1. To change the listening port adjust the port in docker-compose.yml.

Run npm install --legacy-peer-deps to install the nodejs dependencies. Adapt the API_URL in the config.ts if the controller and configuration GUI run on different machines. Afterward run npm run build to create a production build and serve the build/ directory with a webserver of your choice.

Connect to the REST-API server through the frontend of the configuration GUI: http://ip-of-tofino-controller:P4TG_PORT/api

If you update to a newer version, it might be necessary to delete your local storage in your browser for the configuration GUI.

The documentation of the REST-API can be found at the /api/docs endpoint of the REST-API of the controller.