Many Internet service providers (ISPs) have deployed peer-to-peer (P2P) caches in their networks in order to decrease costly inter-ISP traffic. A P2P cache stores parts of the most popular contents locally, and if possible serves the requests of local peers to decrease the inter-ISP traffic. Traditionally, P2P cache resource management focuses on managing the storage resource of the cache so as to maximize the inter-ISP traffic savings. In this paper, we show that when there are many overlays competing for the upload bandwidth of a P2P cache, then in order to maximize the inter-ISP traffic savings, the cache's upload bandwidth should be actively allocated among the overlays. We formulate the problem of P2P cache bandwidth allocation as a Markov decision process and propose three approximations to the optimal cache bandwidth allocation policy. We use extensive simulations and experiments to evaluate the performance of the proposed policies, and show that the bandwidth allocation policy that prioritizes swarms with a small ratio of local peers to all peers in the swarm can improve the inter-ISP traffic savings in BitTorrent-like P2P systems by up to 30%-60%.

Many Internet service providers (ISPs) have deployed peer-to-peer (P2P) caches in their networks in order to decrease costly inter-ISP traffic. A P2P cache stores parts of the most popular contents locally, and if possible serves the requests of local peers to decrease the inter-ISP traffic. Traditionally, P2P cache resource management focuses on managing the storage resource of the cache so as to maximize the inter-ISP traffic savings. In this paper we show that when there are many overlays competing for the upload bandwidth of a P2P cache then in order to maximize the inter-ISP traffic savings the cache's upload bandwidth should be actively allocated among the overlays. We formulate the problem of P2P cache bandwidth allocation as a Markov decision process, and describe two approximations to the optimal cache bandwidth allocation policy. Based on the insights obtained from the approximate policies we propose SRP, a priority-based allocation policy for BitTorrent-like P2P systems. We use extensive simulations to evaluate the performance of the proposed policies, and show that cache bandwidth allocation can improve the inter-ISP traffic savings by up to 30 to 60 percent. We validate the results via BitTorrent experiments on Planet-lab.

Peer-to-peer file-sharing systems are responsible for a significant share of the traffic between Internet service providers (ISPs) in the Internet. In order to decrease their peer-to-peer-related transit traffic costs, many ISPs have deployed caches for peer-to-peer traffic in recent years. We consider how the different types of peer-to-peer caches—caches already available on the market and caches expected to become available in the future—can possibly affect the amount of inter-ISP traffic. We develop a fluid model that captures the effects of the caches on the system dynamics of peer-to-peer networks and show that caches can have adverse effects on the system dynamics depending on the system parameters. We combine the fluid model with a simple model of inter-ISP traffic and show that the impact of caches cannot be accurately assessed without considering the effects of the caches on the system dynamics. We identify scenarios when caching actually leads to increased transit traffic. Motivated by our findings, we propose a proximity-aware peer-selection mechanism that avoids the increase of the transit traffic and improves the cache efficiency. We support the analytical results by extensive simulations and experiments with real BitTorrent clients.

Precongestion notification (PCN) is a packet-marking technique for IP networks to notify egress nodes of a so-called PCN domain whether the traffic rate on some links exceeds certain configurable bounds. This feedback is used by decision points for admission control (AC) to block new flows when the traffic load is already high. PCN-based AC is simpler than other AC methods because interior routers do not need to keep per-flow states. Therefore, it is currently being standardized by the IETF. We discuss various realization options and analyze their performance in the presence of flash crowds or with multipath routing by means of simulation and mathematical modeling. Such situations can be aggravated by insufficient flow aggregation, long round-trip times, on/off traffic, delayed media, inappropriate marker configuration, and smoothed feedback.

P2P applications attract a lot of users and generate the dominant portion of the overall traffic in the Internet today. On the one hand, this large amount of traffic results in high operational costs for ISPs, mainly because of expensive interdomain connections. On the other hand, the performance of P2P applications is constricted by suboptimal peer selection or by bandwidth limitations of ISPs. To overcome these problems, the collaboration of P2P applications and ISPs is desirable, where locality promotion is one of the possible approaches. In this paper, we propose a locality promotion mechanism based on BGP routing information of an ISP and show by simulations that it can reduce inter-domain traffic, prefers shorter connections and peering links over transit links while P2P applications can achieve a better performance as well.

Peer-to-peer file-sharing systems are responsible for a significant share of the traffic between Internet service providers (ISPs) in the Internet. In order to decrease their peer-to-peer related transit traffic costs, many ISPs have deployed caches for peer-to-peer traffic in recent years. We consider how the different types of peer-to-peer caches - caches already available on the market and caches expected to become available in the future - can possibly affect the amount of inter-ISP traffic. We develop a fluid model that captures the effects of the caches on the system dynamics of peer-to-peer networks, and show that caches can have adverse effects on the system dynamics depending on the system parameters. We combine the fluid model with a simple model of inter-ISP traffic and show that the impact of caches cannot be accurately assessed without considering the effects of the caches on the system dynamics. We identify scenarios when caching actually leads to increased transit traffic. Our analytical results are supported by extensive simulations and experiments with real BitTorrent clients.

Locality-awareness is considered as a promising approach to increase the efficiency of content distribution by peer-to-peer (P2P) networks, e.g., BitTorrent. It is intended to reduce the inter-domain traffic which is costly for Internet service providers (ISPs) and simultaneously increase the performance from the viewpoint of the P2P users, i.e, shorten download times. This win-win situation should be achieved by a preferred exchange of information between peers which are located closely to each other in the underlying network topology. A set of studies shows that these approaches can lead to a win-win situation under certain conditions, and to a win-no lose situation in most cases. However, the scenarios used assume mostly homogeneous peer distributions and that all peers have the same access speed. This is not the case in practice according to several measurement studies. Therefore, we extend previous work in this paper by studying scenarios with real-life, skewed peer distributions and heterogeneous access bandwidths of peers. We show that even a win-no lose situation is difficult to achieve under those conditions and that the actual impact for a specific peer depends heavily on the used locality-aware peer selection and the concrete scenario. Therefore, we conclude that current proposals need to be refined so that users of P2P networks can be sure that they also benefit from their use. Otherwise, a broad acceptance of the concept of locality-awareness in the user community of P2P networks will not take place.

Pre-congestion notification (PCN) is a new packet marking scheme based on which simple measurement-based admission control (AC) and flow termination (FT) are implemented. FT is useful for traffic management in unexpected events, e.g., when admitted flows lead to overload on a link after rerouting which may be due to a link or node failure. While AC is a classic flow control function, FT is new and only little understood so far. The limited literature on FT focuses mainly on a single overloaded link. However, when a link or node fails, redirected traffic is likely to cause overload on multiple backup links (bottlenecks) at the same time. As the packet marking probability for flows traversing multiple bottlenecks is larger than for flows traversing only the most severe bottleneck, more traffic is possibly terminated than needed, i.e. over termination occurs. This paper quantifies potential over termination in case of multiple bottlenecks for different FT mechanisms which are currently discussed by the IETF.

Pre-congestion notification (PCN) defines admissible rates (AR) and supportable rates (SR) per link and marks the PCN traffic rate above these thresholds as AR- or SR-overload. The IETF standardizes simple mechanisms for admission control (AC) and flow termination (FT) based on this PCN-feedback for high-priority DiffServ traffic. While admission control (AC) has been extensively discussed in the literature, flow termination (FT) is a new control function. In this paper we propose an algorithm that converts marked AR-overload into marked SR-overload by unmarking appropriate packets. Classic marked flow termination (MFT) is based on marked AR-overload and works well even with a small number of PCN flows per ingress-egress aggregate and in case of multipath routing. Thanks to the new marking converter MFT also works with marked AR-overload so that a single marking scheme suffices to support AC and FT. We investigate whether MFT with marking conversion based on AR-overload retains the benefits classic MFT.

Locality promotion in P2P content distribution networks is currently a major research topic. One of the goals of all discussed approaches is to reduce the interdomain traffic that causes high costs for ISPs. However, the focus of the work in this field is generally on the type of locality information that is provided to the overlay and on the entities that exchange this information. An aspect that is mostly neglected is how this information is used by the peers. In this paper, we consider the predominant approach of Biased Neighbor Selection and compare it with Biased Unchoking, which is an alternative locality aware peer selection strategy that we propose in this paper. We show that both mechanisms complement each other for the BitTorrent file sharing application and achieve the best performance when combined.

Pre-congestion notification (PCN) marks packets when the PCN traffic rate exceeds an admissible link rate and this marking information is used as feedback from the network to take admission decisions for new flows. This idea is currently under standardization in the IETF. Different marking algorithms are discussed and various admission control algorithms are proposed that decide based on the packet markings whether further flows should be accepted or blocked. In this paper, we propose a two-layer architecture that makes the coexistence of various algorithms explicit. We propose novel control algorithms, investigate their behavior under various conditions, and compare them with existing approaches.