Augmented Traffic Control: A tool to simulate network conditions

Overview

Augmented Traffic Control

build-status-image pypi-version

Full documentation for the project is available at http://facebook.github.io/augmented-traffic-control/.

Overview

Augmented Traffic Control (ATC) is a tool to simulate network conditions. It allows controlling the connection that a device has to the internet. Developers can use ATC to test their application across varying network conditions, easily emulating high speed, mobile, and even severely impaired networks. Aspects of the connection that can be controlled include:

  • bandwidth
  • latency
  • packet loss
  • corrupted packets
  • packets ordering

In order to be able to shape the network traffic, ATC must be running on a device that routes the traffic and sees the real IP address of the device, like your network gateway for instance. This also allows any devices that route through ATC to be able to shape their traffic. Traffic can be shaped/unshaped using a web interface allowing any devices with a web browser to use ATC without the need for a client application.

ATC is made of multiple components that interact together:

  • atcd: The ATC daemon which is responsible for setting/unsetting traffic shaping. atcd exposes a Thrift interface to interact with it.
  • django-atc-api: A Django app based on Django Rest Framework that provides a RESTful interface to atcd.
  • django-atc-demo-ui: A Django app that provides a simple Web UI to use atc from a mobile phone.
  • django-atc-profile-storage: A Django app that can be used to save shaping profiles, making it easier to re-use them later without manually re-entering those settings.

By splitting ATC in sub-components, it make it easier to hack on it or build on top of it. While django-atc-demo-ui is shipped as part of ATC's main repository to allow people to be able to use ATC out of the box, by providing a REST API to atcd, it makes it relatively easy to interact with atcd via the command line and opens the path for the community to be able to build creative command line tools, web UI or mobile apps that interact with ATC.

ATC architecture

Requirements

Most requirements are handled automatically by pip, the packaging system used by ATC, and each ATC package may have different requirements and the README.md files of the respective packages should be checked for more details. Anyhow, some requirements apply to the overall codebase:

  • Python 2.7: Currently, ATC is only supported on python version 2.7.
  • Django 1.10: Currently, ATC is only supported using django version 1.10.

Installing ATC

The fact that ATC is splitted in multiple packages allows for multiple deployment scenarii. However, deploying all the packages on the same host is the simplest and most likely fitting most use cases.

To get more details on how to install/configure each packages, please refer to the packages' respective READMEs.

Packages

The easiest way to install ATC is by using pip.

pip install atc_thrift atcd django-atc-api django-atc-demo-ui django-atc-profile-storage

Django

Now that we have all the packages installed, we need to create a new Django project in which we will use our Django app.

django-admin startproject atcui
cd atcui

Now that we have our django project, we need to configure it to use our apps and we need to tell it how to route to our apps.

Open atcui/settings.py and enable the ATC apps by adding to INSTALLED_APPS:

INSTALLED_APPS = (
    ...
    # Django ATC API
    'rest_framework',
    'atc_api',
    # Django ATC Demo UI
    'bootstrap_themes',
    'django_static_jquery',
    'atc_demo_ui',
    # Django ATC Profile Storage
    'atc_profile_storage',
)

Now, open atcui/urls.py and enable routing to the ATC apps by adding the routes to urlpatterns:

...
...
from django.views.generic.base import RedirectView
from django.conf.urls import include

urlpatterns = [
    ...
    # Django ATC API
    url(r'^api/v1/', include('atc_api.urls')),
    # Django ATC Demo UI
    url(r'^atc_demo_ui/', include('atc_demo_ui.urls')),
    # Django ATC profile storage
    url(r'^api/v1/profiles/', include('atc_profile_storage.urls')),
    url(r'^$', RedirectView.as_view(url='/atc_demo_ui/', permanent=False)),
]

Finally, let's update the Django DB:

python manage.py migrate

Running ATC

All require packages should now be installed and configured. We now need to run the daemon and the UI interface. While we will run ATC straight from the command line in this example, you can refer to example sysvinit and upstart scripts.

atcd

atcd modifies network related settings and as such needs to run in privileged mode:

sudo atcd

Supposing eth0 is your interface to connect to the internet and eth1, your interface to connect to your lan, this should just work. If your setting is slightly different, use the command line arguments --atcd-wan and --atcd-lan to adapt to your configuration.

ATC UI

The UI on the other hand is a standard Django Web app and can be run as a normal user. Make sure you are in the directory that was created when you ran django-admin startproject atcui and run:

python manage.py runserver 0.0.0.0:8000

You should now be able to access the web UI at http://localhost:8000

ATC Code Structure

ATC source code is available under the atc directory, it is currently composed of:

  • atc_thrift the thrift interface's library
  • atcd the ATC daemon that runs on the router doing the traffic shaping
  • django-atc-api A django app that provides a RESTful interface to atcd
  • django-atc-demo-ui A django app that provides a simple demo UI leveraging the RESTful API
  • django-atc-profile-storage A django app that allows saving shaping profiles to DB allowing users to select their favorite profile from a list instead of re-entering all the profile details every time.

The chef directory contains 2 chef cookbooks:

  • atc A cookbook to deploy ATC. It also allows to deploy ATC in a Virtual Box VM in order to develop on ATC.
  • atclient Set up a Linux Desktop VM that can be used to test shaping end to end.

atcd

atcd is the daemon that runs on the router that does the shaping. Interaction with the daemon is done using thrift. The interface definition can be found in atc_thrift.thrift.

atc_thrift

atc_thrift defines the thrift interface to communicate with the atcd daemon.

django-atc-api

django-atc-api is a django app that provide a REST API to the atcd daemon. Web applications, command line tools can use the API in order to shape/unshape traffic.

django-atc-demo-ui

django-atc-demo-ui is a simple Web UI to enable/disable traffic shaping. The UI is mostly written in React

django-atc-profile-storage

django-atc-profile-storage allows saving profiles to DB. A typical use case will be to save a list of predefined/often used shaping settings that you want to be able to accessing in just a few clicks/taps.

Developing on ATC

To make ATC development easier, we use Virtual Box and Vagrant to provision and run a VM that will run the ATC daemon and the ATC UI from your git checkout.

Interacting with ATC will only shape the traffic within the VM and not on the host.

Setting up the environment

Note: vagrant is an easy way to set up a test environment, but virtualization will produce different results than a setup on bare-metal. We recommend using vagrant only for testing/development and using bare-metal for setups which require realistic shaping settings.

You will need to install VirtualBox, Vagrant and a couple of plugins:

  • VirtualBox
  • Vagrant
  • Chef DK
  • Install some vagrant plugins:
  • vagrant plugin install vagrant-berkshelf --plugin-version '>= 2.0.1'
  • vagrant plugin install vagrant-omnibus
  • Clone this repo: git clone [email protected]:facebook/augmented-traffic-control.git atc

Running ATC

Once in the repo, go to the chef/atc directory and run:

vagrant up trusty

This will take some time before it completes, once the VM is provision, SSH into it:

vagrant ssh trusty

You should now be able to access ATC at: http://localhost:8080/

Using the Sample Profiles

Once you've got ATC up and running, you can run the script utils/restore-profiles.sh to setup the set of default profiles.

The script needs to be passed a hostname:port with the location of your ATC instance:

utils/restore-profiles.sh localhost:8080

After doing this, you should see the 10 sample profiles listed below in your ATC instance:

  • 2G - Developing Rural
  • 2G - Developing Urban
  • 3G - Average
  • 3G - Good
  • Cable
  • DSL
  • Edge - Average
  • Edge - Good
  • Edge - Lossy
  • No Connectivity

Naturally, you cannot improve your natural network speed by selecting a faster profile than your service. For example, selecting the Cable profile will not make your network faster if your natural connection speed resembles DSL more closely.

Hacking on the code

Hacking on ATC is done from the host and tested in the VM. In order to reflect the changes, you will need to start the services manually.

Both atcd and atcui have their python libraries installed in a python virtualenv so you will need to activate the environment in order to be able to run the services.

The virtualenv is installed in /usr/local/atc/venv/bin/activate .

source /usr/local/atc/venv/bin/activate

Running the daemon

The atcd daemon is running under the root user privileges, all operations below needs to be done as root.

To run the daemon manually, first make sure it is not running in the background:

service atcd stop

And run the daemon:

atcd

Once you are happy with your changes and you want to test them, you will need to kill the daemon and restart it in order to apply the changes.

Running the API/UI

This is a django project and, when running the django built-in HTTP server, will detect code changes and reload automatically.

To run the HTTP REST API and UI:

cd /var/django && python manage.py runserver 0.0.0.0:8000
Owner
Meta Archive
These projects have been archived and are generally unsupported, but are still available to view and use
Meta Archive
Unofficial Tensorflow-Keras implementation of Fastformer based on paper [Fastformer: Additive Attention Can Be All You Need](https://arxiv.org/abs/2108.09084).

Fastformer-Keras Unofficial Tensorflow-Keras implementation of Fastformer based on paper Fastformer: Additive Attention Can Be All You Need. Tensorflo

Yam Peleg 10 Jan 30, 2022
Official Pytorch Implementation of GraphiT

GraphiT: Encoding Graph Structure in Transformers This repository implements GraphiT, described in the following paper: Grégoire Mialon*, Dexiong Chen

Inria Thoth 80 Nov 27, 2022
Opinionated code formatter, just like Python's black code formatter but for Beancount

beancount-black Opinionated code formatter, just like Python's black code formatter but for Beancount Try it out online here Features MIT licensed - b

Launch Platform 16 Oct 11, 2022
An Implicit Function Theorem (IFT) optimizer for bi-level optimizations

iftopt An Implicit Function Theorem (IFT) optimizer for bi-level optimizations. Requirements Python 3.7+ PyTorch 1.x Installation $ pip install git+ht

The Money Shredder Lab 2 Dec 02, 2021
A benchmark for the task of translation suggestion

WeTS: A Benchmark for Translation Suggestion Translation Suggestion (TS), which provides alternatives for specific words or phrases given the entire d

zhyang 55 Dec 24, 2022
Sinkformers: Transformers with Doubly Stochastic Attention

Code for the paper : "Sinkformers: Transformers with Doubly Stochastic Attention" Paper You will find our paper here. Compat This package has been dev

Michael E. Sander 31 Dec 29, 2022
wmctrl ported to Python Ctypes

work in progress wmctrl is a command that can be used to interact with an X Window manager that is compatible with the EWMH/NetWM specification. wmctr

Iyad Ahmed 22 Dec 31, 2022
Official implementation for Multi-Modal Interaction Graph Convolutional Network for Temporal Language Localization in Videos

Multi-modal Interaction Graph Convolutioal Network for Temporal Language Localization in Videos Official implementation for Multi-Modal Interaction Gr

Zongmeng Zhang 15 Oct 18, 2022
Code associated with the paper "Deep Optics for Single-shot High-dynamic-range Imaging"

Deep Optics for Single-shot High-dynamic-range Imaging Code associated with the paper "Deep Optics for Single-shot High-dynamic-range Imaging" CVPR, 2

Stanford Computational Imaging Lab 40 Dec 12, 2022
This is the offical website for paper ''Category-consistent deep network learning for accurate vehicle logo recognition''

The Pytorch Implementation of Category-consistent deep network learning for accurate vehicle logo recognition This is the offical website for paper ''

Wanglong Lu 28 Oct 29, 2022
Code for Active Learning at The ImageNet Scale.

Code for Active Learning at The ImageNet Scale. This repository implements many popular active learning algorithms and allows training with torch's DDP.

Zeyad Emam 47 Dec 12, 2022
Tooling for converting STAC metadata to ODC data model

手语识别 0、使用到的模型 (1). openpose,作者:CMU-Perceptual-Computing-Lab https://github.com/CMU-Perceptual-Computing-Lab/openpose (2). 图像分类classification,作者:Bubbl

Open Data Cube 65 Dec 20, 2022
UniFormer - official implementation of UniFormer

UniFormer This repo is the official implementation of "Uniformer: Unified Transf

SenseTime X-Lab 573 Jan 04, 2023
An unofficial styleguide and best practices summary for PyTorch

A PyTorch Tools, best practices & Styleguide This is not an official style guide for PyTorch. This document summarizes best practices from more than a

IgorSusmelj 1.5k Jan 05, 2023
Expressive Body Capture: 3D Hands, Face, and Body from a Single Image

Expressive Body Capture: 3D Hands, Face, and Body from a Single Image [Project Page] [Paper] [Supp. Mat.] Table of Contents License Description Fittin

Vassilis Choutas 1.3k Jan 07, 2023
The dataset and source code for our paper: "Did You Ask a Good Question? A Cross-Domain Question IntentionClassification Benchmark for Text-to-SQL"

TriageSQL The dataset and source code for our paper: "Did You Ask a Good Question? A Cross-Domain Question Intention Classification Benchmark for Text

Yusen Zhang 22 Nov 09, 2022
VLG-Net: Video-Language Graph Matching Networks for Video Grounding

VLG-Net: Video-Language Graph Matching Networks for Video Grounding Introduction Official repository for VLG-Net: Video-Language Graph Matching Networ

Mattia Soldan 25 Dec 04, 2022
The first machine learning framework that encourages learning ML concepts instead of memorizing class functions.

SeaLion is designed to teach today's aspiring ml-engineers the popular machine learning concepts of today in a way that gives both intuition and ways of application. We do this through concise algori

Anish 324 Dec 27, 2022
Repository for "Improving evidential deep learning via multi-task learning," published in AAAI2022

Improving evidential deep learning via multi task learning It is a repository of AAAI2022 paper, “Improving evidential deep learning via multi-task le

deargen 11 Nov 19, 2022
Similarity-based Gray-box Adversarial Attack Against Deep Face Recognition

Similarity-based Gray-box Adversarial Attack Against Deep Face Recognition Introduction Run attack: SGADV.py Objective function: foolbox/attacks/gradi

1 Jul 18, 2022