Martin Florian

P2P network measurements & monitoring

My team systematically collects empirical data on peer-to-peer networks such as IPFS. Our research uncovers the size and topology of networks, usage patterns, performance bottlenecks as well as the characteristics of commonly stored content.

Lacking a natural vantage point like in centralized services, decentralized networks are inherently hard to get an empirical grip on. Understanding the size, topology and dynamics of such networks is nevertheless highly interesting! Even more so given the surge in popularity that peer-to-peer approaches have recently experienced. Blockchain networks are, in fact, peer-to-peer networks.

Blockchain-based systems aiming for high degrees of decentralization and robustness often require an additional data storage layer, as the blockchain approach is rather unsuitable for storing large amounts of data. Modern decentralized data storage systems like the InterPlanetary Filesystem (IPFS) provide such a layer and are a major focus of our work.

Output: (among others)

  • Ongoing work on the BitSwap layer of IPFS. We discovered that a significant share of the data request traffic in IPFS can be monitored by deploying only a handful of cleverly configured passive nodes. Our monitoring infrastructure continuously produces insights on data usage patterns, popular pieces of content, protocol-related performance characteristics and the size of the network.
  • Work on Mapping the Interplanetary Filesystem. Using our ipfs-crawler, we obtain key insights about the size and structure of the IPFS network. Continuously updated crawling results are available here.
  • Crawler, analysis website and technical report about the MobileCoin network1.

Funding: Weizenbaum, Protocol Labs Research Grant

1 A Federated Byzantine Agreement System.