In tree-based multicast systems, a relatively small number of interior nodes carry the load of forwarding multicast messages. This works well when the interior nodes are highly available, d d cated infrastructure routers but it poses a problem for application-level multicast in peer-to-peer systems. SplitStreamadV esses this problem by striping the content across a forest of interior-nodno# sjoint multicast trees that d stributes the forward ng load among all participating peers. For example, it is possible to construct efficient SplitStream forests in which each peer contributes only as much forwarding bandH d th as it receives. Furthermore, with appropriate content encod ngs, SplitStream is highly robust to failures because a nod e fai ure causes the oss of a single stripe on average. We present thed#' gnand implementation of SplitStream and show experimental results obtained on an Internet testbed and via large-scale network simulation. The results show that SplitStreamd istributes the forward ing load among all peers and can accommod'9 peers with different band0 d capacities while imposing low overhead for forest constructionand maintenance.

This paper addresses the problem of churn---the continuous process of node arrival and departure---in distributed hash tables (DHTs). We argue that DHTs should perform lookups quickly and consistently under churn rates at least as high as those observed in deployed P2P systems such as Kazaa. We then show through experiments on an emulated network that current DHT implementations cannot handle such churn rates. Next, we identify and explore three factors affecting DHT performance under churn: reactive versus periodic failure recovery, message timeout calculation, and proximity neighbor selection. We work in the context of a mature DHT implementation called Bamboo, using the ModelNet network emulator, which models in-network queuing, cross-traffic, and packet loss. These factors are typically missing in earlier simulationbased DHT studies, and we show that careful attention to them in Bamboo's design allows it to function effectively at churn rates at or higher than that observed in P2P file-sharing applications, while using lower maintenance bandwidth than other DHT implementations.

The various proposed DHT routing algorithms embody several di#erent underlying routing geometries. These geometries include hypercubes, rings, tree-like structures, and butterfly networks. In this paper we focus on how these basic geometric approaches a#ect the resilience and proximity properties of DHTs. One factor that distinguishes these geometries is the degree of flexibility they provide in the selection of neighbors and routes. Flexibility is an important factor in achieving good static resilience and e#ective proximity neighbor and route selection. Our basic finding is that, despite our initial preference for more complex geometries, the ring geometry allows the greatest flexibility, and hence achieves the best resilience and proximity performance.

We present an approach to support massively multi-player games on peer-to-peer overlays. Our approach exploits the fact that players in MMGs display locality of interest, and therefore can form self-organizing groups based on their locations in the virtual world. To this end, we have designed scalable mechanisms to distribute the game state to the participating players and to maintain consistency in the face of node failures. The resulting system dynamically scales with the number of online players. It is more flexible and has a lower deployment cost than centralized games servers. We have implemented a simple game we call SimMud, and experimented with up to 4000 players to demonstrate the applicability of this approach.

mechanism to estimate Internet network distance (i.e., round-trip delay or network hops). Network distance estimation is important in many applications, for example, network-aware overlay construction and server selection. There are several proposals for distance estimation in the Internet but they all suffer from problems that limit their benefit. Most rely on a small set of infrastructure nodes that are a single point of failure and limit scalability. Others use sets of peers to compute coordinates but these coordinates can be arbitrarily wrong if one of these peers is malicious. While it may be reasonable to secure a small set of infrastructure nodes, it is unreasonable to secure all peers. PIC addresses these problems: it does not rely on infrastructure nodes and it can compute accurate coordinates even when some peers are malicious. We present PIC's design, experimental evaluation, and an application to network-aware overlay construction and maintenance.

There has been much interest in both unstructured and structured overlays recently. Unstructured overlays, like Gnutella, build a random graph and use flooding or random walks on the graph to discover data stored by overlay nodes. Structured overlays assign keys to data items and build a graph that maps each key to a specific node. The structure of the graph enables efficient discovery of data items given their keys but it does not support complex queries.

Structured peer-to-peer (p2p) overlay networks provide a useful substrate for building distributed applications. They assign object keys to overlay nodes and provide a primitive to route a message to the node responsible for a key. Proximity neighbor selection (PNS) can be used to achieve both low delay routes and low bandwidth usage but it introduces high overhead. This paper presents a detailed evaluation of PNS and heuristic approximations. We describe a new heuristic called constrained gossiping (PNS-CG) and show that it achieves performance similar to perfect PNS with low overhead. We also compare constrained gossiping with previous heuristics and show that it achieves better performance with lower overhead.

We present an application-level implementation of anycast for highly dynamic groups. The implementation can handle group sizes varying from one to the whole Internet, and membership maintenance is e#cient enough to allow members to join for the purpose of receiving a single message. Key to this e#ciency is the use of a proximity-aware peer-to-peer overlay network for decentralized, lightweight group maintenance; nodes join the overlay once and can join and leave many groups many times to amortize the cost of maintaining the overlay. An anycast implementation with these properties provides a key building block for distributed applications. In particular, it enables management and location of dynamic resources in large scale peer-to-peer systems. We present several resource management applications that are enabled by our implementation.

In this paper we motivate a Context Fusion Network (CFN), an infrastructure model that allows contextaware applications to select distributed data sources and compose them with customized data-fusion operators into a directed acyclic information fusion graph. Such a graph represents how an application computes high-level understandings of its execution context from low-level sensory data. Multiple graphs by different applications inter-connect with each other to form a global graph. A key advantage of a CFN is re-usability, both at code-level and instance-level, facilitated by operator composition. We designed and implemented a distributed CFN system, Solar, which maps the logical operator graph representation onto a set of overlay hosts. In particular, Solar meets the challenges inherent to heterogeneous and volatile ubicomp environments. By abstracting most complexities into the infrastructure, Solar facilitates both the development and deployment of context-aware applications. We present the operator composition model, basic services of the Solar overlay network, and programming support for the developers. We also discuss some applications built with Solar and the lessons we learned from our experience. 1

This paper presents Araneola 1, a scalable reliable application-level multicast system for highly dynamic wide-area environments. Araneola supports multi-point to multi-point reliable communication in a fully distributed manner while incurring constant load (in terms of message and space complexity) on each node. For a tunable parameter k ≥ 3, Araneola constructs and dynamically maintains a basic overlay structure in which each node’s degree is either k or k +1, and roughly 90 % of the nodes have degree k. Empirical evaluation shows that Araneola’s basic overlay achieves three important mathematical properties of k-regular random graphs (i.e., random graphs in which each node has exactly k neighbors) with N nodes: (i) its diameter grows logarithmically with N; (ii) it is generally k-connected; and (iii) it remains highly connected following random removal of linear-size subsets of edges or nodes. The overlay is constructed and maintained at a low cost: each join, leave, or failure is handled locally, and entails the sending of only about 3k messages in total, independent of N. Moreover, this cost decreases as the churn rate increases. The low degree of Araneola’s basic overlay structure allows for allocating plenty of additional bandwidth for specific application needs. In this paper, we give an example for such a need — communicating with nearby nodes; we enhance the basic overlay with additional links chosen according to geographic