Peer-to-peer streaming systems are becoming highly popular for IP Television (IPTV).'RsIost systems can be categorized as either tree-based or mesh-based, and as either pushbased or pull-based. However, there is a lack of clear understanding of how these different mechanisms perform comparatively in a real-world setting. In this paper, we compare two representative streaming systems using mesh-based and multiple tree-based overlay routing through deployments on the PlanetLab wide-;area experimentation platform. To the best of our knowledge, this is the first study to directly compare streaming overlay architectures in real Internet settings. Our results indicate that mesh-based systems inject a much higher number of duplicate packets into the network, but they perform better under a variety of conditions. In particular, mesh-based systems give consistently higher application goodput when the number of overlay nodes, or the streaming rates increase. They also perform better under churn and large flash crowds. Their performance suffers when latencies among peers are high, however. Overall, mesh-based systems appear to be a better choice than multi-tree based systems for peer-to-peer streaming at a large scale.

Abstract. 1The arms race between copyright agencies and P2P users is an ongoing and evolving struggle. On the one hand, content providers are using several techniques to stealthily find unauthorized distribution of copyrighted work in order to deal with the problem of Internet piracy. On the other hand, P2P users are relying increasingly on blacklists and anonymization methods in order to avoid detection. In this work, we propose a number of techniques to reveal copyright monitors' current approaches and evaluate their effectiveness. We apply these techniques on data we collected from more than 2.75 million BitTorrent swarms containing 71 million IP addresses. We provide strong evidence that certain nodes are indeed copyright monitors, show that monitoring is a world-wide phenomenon, and devise a methodology for generating blacklists for paranoid and conservative P2P users.

We conduct a detailed simulation study to examine the impact of localizing P2P traffic within network boundaries on an ISP's profitability. A distinguishing aspect of our work is the focus on Internet-wide implications, i.e., how adoption of localization within an ISP affects both itself and other ISPs. Our simulations are based on detailed models of inter-AS P2P traffic and inter-AS routing, localization models that can predict the extent to which P2P traffic is reduced, and pricing models to predict the impact of changes in traffic on an ISP's profit. To evaluate our models we use a large-scale crawl of BitTorrent involving over 138 million users sharing 2.75 million files. Our results show that the benefits of localization must not be taken for granted. Some of our key findings include: (i) residential ISPs can actually lose money when localization is employed and some will not see increased profitability until other ISPs employ localization; (ii) the reduction in costs due to localization will be limited for small ISPs and tends to grow only logarithmically with client population; and (iii) some ISPs can better increase profitability through alternate strategies to localization by taking advantage of the business relationships they have with others.

No abstract

Abstract-The goal of enabling ubiquitous video broadcasting on the Internet has been a long cherished vision in the networking community. Prior efforts aimed at achieving this goal based on the IP Multicast architecture have been unsuccessful. In recent years, peer-to-peer (P2P) streaming has emerged as a promising alternative technology, which has matured to the point that there are several commercial offerings available to users. While these developments are encouraging, P2P streaming systems are susceptible to attacks by malicious participants, and their viability depends on how effectively they can perform under such attacks. In this paper, we explore this issue in the context of meshbased designs, which have emerged as the dominant architecture for P2P streaming.We provide a taxonomy of the implicit commitments made by nodes when peering with others. We show that when these commitments are not enforced explicitly, they can be exploited by malicious nodes to conduct attacks that degrade the data delivery service. We consider an important class of attacks where malicious nodes deliberately become neighbors of a large number of nodes and do not upload data to them. We focus on these attacks given the limited attention paid to them, and the significant impact they can have on overall data delivery. We present mechanisms that can enhance the resilience of meshbased streaming against such attacks. A key part of the solution is a novel reputation scheme that combines feedback from both the control and data planes of the overlay. We evaluate our design with real-world experiments on the PlanetLab testbed and show that our design is effective. Even when there are 30% attackers, nodes can receive 92% of the data with our schemes compared to 10% of the data without our schemes. Overall these results indicate the feasibility of enabling effective P2P streaming even under the presence of malicious participants.