Skip to content
This repository has been archived by the owner on Dec 6, 2021. It is now read-only.

dioptra-io/multilevel-mda-lite

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Multilevel MDA-Lite Paris Traceroute (MMLPT)

Multilevel MDA-Lite Paris Traceroute is a portable (Linux, MacOS, Windows) traceroute-like tool capable of both giving IP and router level view. It supports IPv4 for IP level and router level, and IPv6 only for IP level.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Prerequisites

MMLPT is written in Python, and uses different libraries to work.

To craft, send and parse packets, it uses scapy:

sudo pip install scapy

If you are on OSX, scapy does not work natively and you must install libdnet and its python wrapper. It works for python2+.

For python3+, libdnet has no python wrapper at the moment, so we recommend you using kamene, which is a fork of scapy supporting python3+.

We will remove the dependency to kamene as soon as libdnet has a wrapper for python3 working with scapy.

sudo pip install kamene

To provide a rich graph view, and to compute the IP and router level graphs, MMLPT uses graph-tool, which is a boost C++ library with a python wrapper.

Follow the instructions in the following link to install graph-tool. As the C++ Library makes an extensive use of metaprogramming and templates, it can take a while (up to 30 minutes on a recent laptop) to compile graph-tool:

Graph-tool

Installing

To install then, just tap:

sudo python setup.py install

Running

To run a simple MDA-Lite traceroute at IP level, tap

sudo python MDALite.py destination

To run MDA-Lite traceroute at both IP and router level, tap:

sudo python MDALite.py -a destination

You can tap the following command for the different options

sudo python MDALite.py --help

Recall that on very large topologies, the measurement can take few minutes, up to 10-15 on very complex topologies.

Exploring traceroute results

You should have this kind of output:

(0) : 127.0.0.1 -> ['192.168.0.254']

(1) : 192.168.0.254 -> ['137.194.164.254']

(2) : 137.194.164.254 -> ['137.194.4.240']

(3) : 137.194.4.240 -> ['212.73.200.45']

(4) : 212.73.200.45 -> ['4.69.161.114', '4.69.161.110']

(5) : 4.69.161.114 -> ['4.68.111.194']

4.69.161.110 -> ['4.68.111.194']

(6) : 4.68.111.194 -> ['62.115.122.10', '62.115.122.4']

(7) : 62.115.122.10 -> ['62.115.122.139']

62.115.122.4 -> ['62.115.123.12']

(8) : 62.115.123.12 -> ['62.115.138.236']

62.115.122.139 -> ['62.115.138.104']

(9) : 62.115.138.104 -> ['62.115.133.31', '62.115.136.37', '62.115.141.205', '62.115.119.115', '62.115.136.113', '62.115.139.107', '62.115.114.171', '62.115.136.35']

62.115.138.236 -> ['62.115.136.107', '62.115.133.29', '62.115.136.21', '62.115.136.23', '62.115.114.167', '62.115.141.201', '62.115.119.113', '62.115.136.25']

(10) : 62.115.136.107 -> ['213.248.79.106']

62.115.133.31 -> ['213.248.79.106']

62.115.133.29 -> ['213.248.79.106']

62.115.136.37 -> ['213.248.79.106']

62.115.141.205 -> ['213.248.79.106']

62.115.136.21 -> ['213.248.79.106']

62.115.119.115 -> ['213.248.79.106']

62.115.136.23 -> ['213.248.79.106']

62.115.136.113 -> ['213.248.79.106']

62.115.139.107 -> ['213.248.79.106']

62.115.114.171 -> ['213.248.79.106']

62.115.114.167 -> ['213.248.79.106']

62.115.141.201 -> ['213.248.79.106']

62.115.119.113 -> ['213.248.79.106']

62.115.136.25 -> ['213.248.79.106']

62.115.136.35 -> ['213.248.79.106']

(11) : 213.248.79.106 -> ['195.12.233.115']

(12) : 195.12.233.115

Routers found :

['62.115.136.113', '62.115.136.107']

['62.115.136.23', '62.115.133.31', '62.115.141.201', '62.115.136.21', '62.115.136.25']

['62.115.133.29', '62.115.119.113', '62.115.119.115']

['62.115.136.35', '62.115.136.37']

If the text output is not verbose enough, you can ask for a serialized version of the graph that basically keeps all the informations of the traceroute with the -o option

sudo python MDALite.py -o traceroute.xml -a destination 
from graph_tool import *
g = load_graph("/path/to/serialized/graphs.xml")

# Manipulate the graph via its properties.
v_properties = g.vertex_properties
e_properties = g.edge_properties
g_properties = g.graph_properties

# Example: ip_ids, print ip_ids of an ip_address
ip_address = g.vertex_properties["ip_address"]
ip_ids = g.vertex_properties["ip_ids"]
for v in g.vertices():
    if ip_address[v] == "an_ip_of_your_choice_in_the_traceroute":
        print ip_ids[v]

Deploying in a docker container.

Dockerfile provides an example of a Docker script that allows you to deploy MMLPT in a Dockercontainer (in the example with an ubuntu image). To build the container, launch the command:

docker build --build-arg "CACHE_DATE=$(date)" -t "name/of/your/docker/image" . 

Then push it to docker repository:

docker push name/of/your/docker/image

To run it:

docker run -d name/of/your/docker/image