摘要
A peer-to-peer （P2P） network is a distributed application architecture which provides many attractive features, such as availability, self-organization, load-balancing, and anonymity. However, P2P network has created significant problems to network operators by generating large volumes of inter autonomous system （inter-AS） traffic. Focusing on the BitTorrent swarming protocol, this paper proposes an approach which aims to reduce P2P generated inter-AS traffic. In particular, the approach can reduce inter-AS traffic by 50% to 70%. Moreover, it can improve the downloading speed by 60% for the popular torrents. The evaluation shows that controlled regional-based contents replication can effectively achieve this goal. Furthermore, the approach is incrementally deployable. Network regions in which the system gets deployed can solve their P2P generated inter-AS traffic problems autonomously, i.e., without any Internet service providers-collaboration and any requirement, the system can be deployed in the entire Internet.
A peer-to-peer （P2P） network is a distributed application architecture which provides many attractive features, such as availability, self-organization, load-balancing, and anonymity. However, P2P network has created significant problems to network operators by generating large volumes of inter autonomous system （inter-AS） traffic. Focusing on the BitTorrent swarming protocol, this paper proposes an approach which aims to reduce P2P generated inter-AS traffic. In particular, the approach can reduce inter-AS traffic by 50% to 70%. Moreover, it can improve the downloading speed by 60% for the popular torrents. The evaluation shows that controlled regional-based contents replication can effectively achieve this goal. Furthermore, the approach is incrementally deployable. Network regions in which the system gets deployed can solve their P2P generated inter-AS traffic problems autonomously, i.e., without any Internet service providers-collaboration and any requirement, the system can be deployed in the entire Internet.
基金
supported by the National Natural Science Foundation of China under Grant No. 61001084
