This repository contains all the parsing files and wrapping scripts used for the OVH Weather dataset. It is linked to the paper Revealing the Evolution of a Cloud Provider Through its Network Weather Map, accepted at the Internet Measure Conference 2022 (IMC'22).
Additionally, the repository includes all the analysis scripts used to generate the plots from the paper, as well as scripts to read the YAML files.
We provide the weathermap_parse.py Python script. This script reads SVG files from a directory and parses them in a user-friendly YAML format. As explained in the paper, several sanity checks are performed to ensure that the SVG->YAML translation is correct.
To parse the OVH weather map SVG files from a directory, one can use the parsing script as follows.
weathermap_parse.py <path-to-directory>
This will parse all files from the directory given as argument. The YAML files are located in a new directory <path-to-directory>_yaml
. For example, if the argument is ~/SomeUser/SomeOVHData
, the parsed YAML files will be located in ~/SomeUser/SomeOVHData_yaml
.
The YAML parsed files can be loaded in memory for further analysis. As a starting point, we implemented YAML readers for the OVH Weather dataset in the following languages (available in the yaml_readers directory):
Any contribution to extend this list is welcomed.
The scripts used for the (basic) analysis of the OVH network based on the YAML files are available in the ovh-parsing
directory. More information can be found in the related README
.
All the analyzing scripts are written in Rust and require the Cargo toolchain. The output files from the analysis are located in the csv
directory. The plots
directory uses theses files to generates the Figures from the paper (which are located in figures
).
@inproceedings{piraux2022revealing,
title={Revealing the evolution of a cloud provider through its network weather map},
author={Piraux, Maxime and Navarre, Louis and Rybowski, Nicolas and Bonaventure, Olivier and Donnet, Benoit},
booktitle={Proceedings of the 22nd ACM Internet Measurement Conference},
pages={298--304},
year={2022}
}