Scalable distribution of large files has been the area of much research and commercial interest in the past few years. In this paper, we describe the CoBlitz system, which efficiently distributes large files using a content distribution network (CDN) designed for HTTP. As a result, CoBlitz is able to serve large files without requiring any modifications to standard Web servers and clients, making it an interesting option both for end users as well as infrastructure services. Over the 18 months that CoBlitz and its partner service, CoDeploy, have been running on PlanetLab, we have had the opportunity to observe its algorithms in practice, and to evolve its design. These changes stem not only from observations on its use, but also from a better understanding of their behavior in real-world conditions. This utilitarian approach has led us to better understand the effects of scale, peering policies, replication behavior, and congestion, giving us new insights into how to better improve their performance. With these changes, CoBlitz is able to deliver in excess of 1 Gbps on PlanetLab, and to outperform a range of systems, including research systems as well as the widely-used BitTorrent. 1

This paper addresses the problem of fault resilience of overlay-based live media streaming from two aspects: (1) how to construct a stable multicast tree that minimizes the negative impact of frequent member departures on existing overlay, and (2) how to efficiently recover from packet errors caused by end-system or network failures. In particular, this paper makes two contributions: (1) A distributed Reliability-Oriented Switching Tree (ROST) algorithm that minimizes the failure correlation among tree nodes. By exploiting both bandwidth and time properties, the algorithm constructs a more reliable multicast tree than existing algorithms that solely minimize tree depth, while not compromising the quality of the tree in terms of service delay and incurring only a small protocol overhead; (2) A simple Cooperative Error Recovery (CER) protocol that helps recover from packet errors efficiently. Recognizing that a single recovery source is usually incapable of providing timely delivery of the lost data, the protocol recovers from data outages using the residual bandwidths from multiple sources, which are identified using a minimum-losscorrelation algorithm. Extensive simulations are conducted to demonstrate the effectiveness of the proposed schemes.

Even though overlay streaming is an inherently fault tolerant and stable system architecture, careful neighbour selection is a significant task. Inappropriate routing decisions can lead to an unstable topology with only a few very important nodes on which a large set of succeeding nodes depend. The presented algorithm selects streaming neighbours based on local information, passing knowledge to parent nodes only. Similar to SplitStream [6], it creates inner-node disjoint multicast trees. The created topologies are broad and have short paths, thus improving the resistance to node failure and intentional attacks. A malicious node can neither gain any knowledge about di#erent regions of the topology other than its own successors nor deliberately move to a more important position in the hierarchy. The characteristics of the created topologies are revised in a static simulation study calculating the vertex connectivity and packet loss on node disconnections.

Among the proposed overlay multicast protocols, treebased systems have proven to be highly scalable and efficient in terms of physical link stress and end-to-end latency. Conventional tree-based protocols, however, distribute the forwarding load unevenly among the participating peers. An effective approach for addressing this problem is to stripe the multicast content across a forest of disjoint trees, evenly sharing the forwarding responsibility among participants. DHTs seem to be naturally well suited for the task, as they are able to leverage the inherent properties of their routing model in building such a forest. In heterogeneous environments, though, DHT-based schemes for tree (and forest) construction may yield deep, unbalanced structures with potentially large delivery latencies. This paper introduces Magellan, a new overlay multicast protocol we have built to explore the tradeoff between fairness and performance in these environments. Magellan builds a data-distribution forest out of multiple performance-centric, balanced trees. It assigns every peer in the system a primary tree with priority over the peer’s resources. The peers ’ spare resources are then made available to secondary trees. In this manner, Magellan achieves fairness, ensuring that every participating peer contributes resources to the system. By employing a balanced distribution tree with O(lg N)-bounded, end-to-end hop-distance, Magellan also provides high delivery ratio with comparable low latency. Preliminary simulation results show the advantage of this approach.

This paper describes a simple scheduling procedure for use in multicast data distribution within a logistical networking infrastructure. The goal of our scheduler is to generate a distribution schedule that will exploit the best network paths by using historic network performance information. A ”spanning tree ” is constructed between available logistical depots to help reduce the overall time of data movement. Our hypothesis is that we can generate appropriate schedules from historical network measurements. In order to evaluate the scheduling procedure we have employed the multicast operation used in the Internet Backplane Protocol (IBP) middleware suite. Investigation into the merits of such a scheduling procedure involved a control group that performs a broadcast to a set of logistical depots and an experimental group that is configured to perform a multicast via schedules generated based on historical network data. All testing was conducted on PlanetLab, a distributed network service testbed. 1

Abstract — One of the most important challenges of selforganized, overlay systems for large-scale group communication lies in these systems ability to handle the high degree of transiency inherent to their environment. While a number of resilient protocols and techniques have been recently proposed, achieving high delivery ratios without sacrificing end-to-end latencies or incurring significant additional costs has proven to be a difficult task. In this paper we review some of these approaches and experimentally evaluate their effectiveness by contrasting their performance and associated cost through simulation and widearea experimentation. I.

Resilience to failures and deliberate attacks is becoming an essential requirement in most communication networks today. This also applies to P2P Overlays which on the one hand are created on top of communication infrastructures, and therefore are equally affected by failures of the underlying infrastructure, but which on the other hand introduce new possibilities like the creation of arbitrary links within the overlay. In this article, we present a survey of strategies to improve resilience in communication networks as well as in P2P overlay networks. Furthermore, our intention is to point out differences and similarities in the resilience-enhancing measures for both types of networks. By revising some basic concepts from graph theory, we show that many concepts for communication networks are based on well-known graph-theoretical problems. Especially, some methods for the construction of protection paths in advance of a failure are based on very hard problems, indeed many of them are in NP and can only be solved heuristically or on certain topologies. P2P overlay networks evidently benefit from resilienceenhancing strategies in the underlying communication infrastructure, but beyond that, their specific properties pose the need for more sophisticated mechanisms. The dynamic nature of peers requires to take some precautions, like estimating the reliability of peers, redundantly storing information, and provisioning a reliable routing. 1

Abstract not found