Impacto de redes sociais na difusão de conteúdo em swarms Bittorrent

Abstract

Peer-to-peer applications are very important on Internet since last decade. Its success as a file-sharing application is unquestionable and has been repeated on applications such cloud computing, social network and video broadcasting. However, it is notorious that the traffic generated by peer-to-peer is not AS friendly: peers on a P2P system are usually geographically sparse, making more expensive the data traffic. Because of it, some content providers perform traffic shaping on their network, reducing P2P packet priority and, by consequence, its performance. Yet, other important problem on P2P systems is the freeriding behavior, where a peer doesn’t contribute to the system with its resources. Today, there are many proposed approaches to reduce inter-AS traffic and freeriding by finding a way to provide long term relationship between peers. With the present success of social networks, some proposals intend to investigate if social contexts can alleviate the impact of freeriding and inter-AS traffic. Through measurements performed in a Bittorrent real-world swarm data-set, we present a characterization that intends to investigate the influences of swarm social dissemination on Facebook in peer relationship and also performance metrics, location, and freeriding reduction. Results achieved by us shows that using social dissemination provides a notable improvement on solving the freeriding problem and peer geographic dispersion. So, a protocol that uses a geographic or social aware peer selection can have better performance on social disseminated swarms.