We present Tapestry, a peer-to-peer overlay routing infrastructure offering efficient, scalable, locationindependent routing of messages directly to nearby copies of an object or service using only localized resources. Tapestry supports a generic Decentralized Object Location and Routing (DOLR) API using a self-repairing, softstate based routing layer. This paper presents the Tapestry architecture, algorithms, and implementation. It explores the behavior of a Tapestry deployment on PlanetLab, a global testbed of approximately 100 machines. Experimental results show that Tapestry exhibits stable behavior and performance as an overlay, despite the instability of the underlying network layers. Several widely-distributed applications have been implemented on Tapestry, illustrating its utility as a deployment infrastructure.

Broose is a peer-to-peer protocol based on the De-Bruijn topology allowing a distributed hashtable to be maintained in a loose manner. Each association is stored on k nodes to allow higher reliability with regard to node failures. Redundancy is also used when storing contacts avoiding complex topology maintenance for node departures and arrivals. It uses a constant size routing table of O(k) contacts for allowing lookups in O(log N) message exchange (where N is the number of nodes participating). It can also be parameterized for obtaining O(log N / log log N) steps lookups with a routing table of size O(k log N). These bounds hold with high probability. Moreover, the protocol allows load balancing of hotspots of requests for a given key as well as hotspots of key collisions. The goal is to obtain a protocol as practical as Kademlia based on the De-Bruijn topology.

Abstract: In this paper, we consider the problem of multicasting a stream of packets in a large scale peer-to-peer environment. In that context peers should have incentive to cooperate. We present PrefixStream, an algorithm that addresses this problem by using reciprocity in packet forwarding. Each node thus has incentive to forward since recipients send back other packets of the stream. To achieve this efficiently, PrefixStream strips the content across two sets of clustered trees built upon the symmetric de Bruijn graph. This both allows to banish nodes that do not respect reciprocity of exchanges and gives resilience to node failures. Furthermore, it reduces the forwarding load of every node to the stream bandwidth (every node uploads as much as it downloads) even when the size of its cluster varies. Conversely to previously proposed hierarchical schemes, PrefixStream promotes disjoint clustering. This enables loose maintenance and network latencies optimization. We sketch the design of PrefixStream and analyze its performances.

Distributed Hash Tables are the core technology on a significant share of system designs for Peer-to-Peer information sharing. Typically, a location mechanism is provided and object identifiers act as keys in the index of object locations. When introducing a search mechanism, where single words are used as keys, the key image cardinality will be driven by the word popularity and most of the present designs will be unable to load balance the index among the nodes. We present two contributions: A design that allows participating nodes to load balance the indexing of popular keys and avoid content hot-spots on single nodes; A distributed mechanism for probabilistic filtering of popular keys (with low search relevance) that paves the way for scalable full content indexing.

Abstract. In Peer to Peer (P2P) systems, peers can join and leave the network whenever they want. Such “freedom ” causes unpredictable network environment which leads to the most complex design challenge of a p2p protocol: how to make p2p service available under churn? What is more, where is the extreme of a system’s resistibility to high churn? A careful evaluation of some typical peer-to-peer networks will contribute a lot to choosing, using and designing a certain kind of protocol in special applications. In this paper we analyze the performance of Chord [1], Tapestry [2], Kelips [3], Kademlia [4] and Koorde [5], then find out the crash point [6] of each network based on the simulation experiment. Finally, we propose a novel way to help nodes survive under high churn.

We are building Coral, a peer-to-peer content distribution system. Coral creates self-organizing clusters of nodes that fetch information from each other to avoid communicating with more distant or heavily-loaded servers. Coral indexes data, but does not store it. The actual content resides where it is used, such as in nodes' local web caches. Thus, replication happens exactly in proportion to demand.

Abstract not found