Understanding long-term trends of DHT networks

Distributed hash tables (DHTs) have been prevalent within the Internet community for many years. Crawling a DHT network allows researchers to monitor the behavior of peers, determine their geographic location, and predict their trend, etc. To characterize a DHT network accurately and comprehensively, the measurements should be conducted in the long term, which are rarely considered in the former studies. In this paper, we crawl the snapshots of two biggest DHT networks, Mainline and Kad, in 2009 and 2016 respectively to find out their changes and characters. Firstly, we use the hybrid approach proposed in our former studies to take snapshots for Mainline and Kad. Then, we characterize DHT networks regarding metrics such as size variety, peer liveness, ID liveness, and geography transformation during the past seven years. Finally, with the data and model, we analyze several interesting features and changes of these two DHT networks, which would be useful for the design and measurement of the DHT network in the future.