Active measurements on BitTorrent and eMule ecosystem over the Internet

摘要

Peer-to-peer applications have generated most volume of Internet traffic, which putting unprecedented pressure on ISPs. Besides simple throttling the traffic, some proposals suggested to optimize the overlay construction using locality-aware mechanism. Thus, detecting the individual peers, analyzing their distribution becomes a challenging task, which has not received the deserved attention. In this paper, we demonstrate an active measurement methodology to continuously trace the real-world BitTorrent and eMule/eDonkey peers over the Internet from a stub network and for a long period of time. Our measurements achieve the ability of real-time scanning the online peers, avoiding the potential copyright infringement and heavy traffic on Internet. During the over eight-month measurements, from July 15 2010 to March 31 2011, we crawled 33,008 torrents for BitTorrent and 19,840 meta-info for eMule/eDonkey, collected 8,963,776 and 1,538,408 distinct peers separately. And we thoroughly analyzed some distributions and preferences of peers, such as the geographical position, organization, type of IP address and the most used port, etc.