Day in the Life of RIPE Atlas: Operational Insights and Applications in Network Measurements

TL;DR

This paper provides an in-depth analysis of RIPE Atlas's daily operations, measurement contributions, and potential biases, offering insights and recommendations for researchers utilizing this global network measurement platform.

Contribution

It offers a comprehensive operational analysis of RIPE Atlas, highlighting measurement patterns, biases, and applications, along with guidelines for ethical and reproducible research.

Findings

Most results are from built-ins and anchor meshes (89%)

Analysis of measurement bias and probe contributions

Applications in censorship detection and network topology insights

Abstract

Network measurement platforms are increasingly popular among researchers and operators alike due to their distributed nature, simplifying measuring the remote parts of the Internet. RIPE Atlas boasts over 12.9K vantage points in 178 countries worldwide and serves as a vital tool for analyzing anycast deployment, network latency, and topology, to name a few. Despite generating over a terabyte of measurement results per day, there is limited understanding of the underlying processes. This paper delves into one day in the life of RIPE Atlas, encompassing 50.9K unique measurements and over 1.3 billion results. While most daily measurements are user-defined, it is built-ins and anchor meshes that account for 89% of produced results. We extensively examine how different probes and measurements contribute to the daily operations of RIPE Atlas and consider any bias they may introduce. Furthermore, we demonstrate how existing measurements can be leveraged to investigate censorship, traceroute symmetry, and the usage of reserved address blocks, among others. Finally, we curate a set of recommendations for researchers using the RIPE Atlas platform to foster transparency, reproducibility, and ethics.