Toward an accurate snapshot of DHT networks

DHTs have been widely used in designing various applications, especially P2P file sharing systems. It is crucial but difficult to get an accurate snapshot of a DHT network, since such network is serverless and peers join and depart dynamically. In this letter, we analyze and empirically verify that neither breadth-first search nor depth-first search is optimal for DHT snapshot crawling, due to uneven ID distribution of contacts in the routing table of each DHT peer. We then propose a hybrid search strategy which could both start up quickly and slow down slowly. The experiment performed on Mainline shows that this approach significantly outperforms breadth-first or depth-first search strategy.