No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Overlay Design Mechanisms for Heterogeneous, Large-Scale, Dynamic P2P Systems
Abstract
Large-scale, heterogeneous peer-to-peer (P2P) systems impose a set of diverse requirements. Current solutions do commonly only address a subset of these requirements since there are a number of trade-offs and constraints due to the different dimensions and aims they address. We present a novel approach for designing overlay networks for large-scale, highly dynamic, and heterogeneous P2P systems. A set of mechanisms is proposed to meet the complete set of requirements while keeping the trade-offs and constraints in balance. To handle effectively the large number of peers, they are clustered in manageable groups considering the requirements on their stability. The novelty in this approach is in the identification of the core services and operations of the aforementioned systems. On the basis of the requirements of those services and operations, peers are assigned the most suitable roles. Role relationships are further introduced to enable (and provide) incentives for the peers to adopt the most suitable roles while selecting an efficient overlay structure to preserve efficiency, robustness, and scalability. The proposed set of mechanisms is realized in Omicron, a novel hybrid P2P approach.
... The tool utilizes the aforementioned rich role model and separates the provided functionality in three layers (user, overlay network and underlay network). Chord [SMLN + 03], Gnutella [PSAS01], JXTA [TAA + 03] and Omicron [DMS04] have been effectively ported into our simulator. 6. Performance evaluation. ...
... Derived common message structure (cf. [DMS04], use proactive mechanisms to provide efficient indexing functionality for advertised resources. The majority of their implementations provide theoretical upper bounds on the communication cost in worst case scenarios. ...
... On the other hand, a richer and better-balanced role model has been proposed for Omicron [DMS04]. Here four different core roles have been identified: the Router, the Cacher, the Indexer and the Maintainer. ...
... In many use cases, the members of an overlay network form subgroups at run-time, e.g. for better efficiency, resilience or security. This may include multicast groups, semantic clusters or functional clusters (like in Omicron [DMS04]). These subgroups will most likely have different requirements than the "global" overlay (after all, that is why they are formed in the first place). ...
... Note that due to the structure of these graphs, a functional overlay implementation will need more than one view here to compensate for empty buckets if the ID space is not completely filled up with nodes. Such approaches are described in [LKRG03,DMS04]. ...
... Darlagiannis and others recently presented Omicron [DMS04], a design study for structured overlay networks. The idea is to split the algorithm that each overlay member traditionally executes into a number of simpler services that have different requirements. ...
... Overlay networks have been recommended for self-organizing networks [23], a critical feature of P2P infrastructure. Overlay networks have been identified to be effective design mechanisms for deploying heterogeneous, large-scale, P2P systems [24]. The purpose of P2P-based overlay networks is that they abstract the complicated connectivity of a P2P network to higher-level programmatical view of the peers that make up the network [18]. ...
... 4) Stabilization: Ad-hoc infrastructure needs to support variable churn-rates that may result if an existing peer goes down or a new node joins the network. P2P routing algorithms are designed to stabilize the overall network against such random changes [24]. Further, such algorithms are also designed to redistribute resources that are placed all over the network. ...
... P2P algorithms, in this case, will redistribute the information over the network. Stabilization is an important way of providing fault-tolerance towards the overall network [24]. 5) Routing: Every peer in a P2P network stores routing information of nearby peers that can be reached. ...
... Locality can be defined as the proximity of the hosts involved in a network operation [6]. This proximity can be measured using different criteria, such as the physical location, the number of hops between hosts, the link latency, among others. ...
... Intuitively, this is also the property that enables a system to limit the impact of local operations on the wide-area performance, both during regular operation and under fault conditions [7]; 2) Relative Delay Penalty (RDP): it is can be measured through the additional packet delay introduced by the overlay when some message is exchanged between two nodes. It is defined as the ratio of the latency experienced when sending data using the overlay to the latency experienced when the same data is sent directly (i.e., using the underlying network) [6]; 3) Link stress: quantifies the usage of the underlying network by the overlay. It is defined as the number of tunnels that send traffic over a physical link [6]; 4) Control Overhead: used when analyzing performance, it can be measured using the time consumed by the joining and by the recovery processes [6]; 5) Quality of Data Path: also deployed when analyzing performance, it evaluates the network proximity by means of the End-to-End Delay (EED) and the Absolute Delay Penalty (ADP) in the overlay multicast tree. ...
... It is defined as the ratio of the latency experienced when sending data using the overlay to the latency experienced when the same data is sent directly (i.e., using the underlying network) [6]; 3) Link stress: quantifies the usage of the underlying network by the overlay. It is defined as the number of tunnels that send traffic over a physical link [6]; 4) Control Overhead: used when analyzing performance, it can be measured using the time consumed by the joining and by the recovery processes [6]; 5) Quality of Data Path: also deployed when analyzing performance, it evaluates the network proximity by means of the End-to-End Delay (EED) and the Absolute Delay Penalty (ADP) in the overlay multicast tree. EED is the Round-Trip Time (RTT) between a parent and a child node in the overlay multicast tree. ...
The continuous growth of peer-to-peer networks has made them responsible for a considerable portion of the current Internet traffic. For this reason, improvements in P2P network resources usage are of central importance. One effective approach for addressing this issue is the deployment of locality algorithms, which allow the system to optimize the peers selection policy for different network situations and, thus, maximize performance. To date, several locality algorithms have been proposed for use in P2P networks. However, they usually adopt heterogeneous criteria for measuring the proximity between peers, which hinders a coherent comparison between the different solutions. In this paper, we develop a thoroughly review of popular locality algorithms, based on three main characteristics: the adopted network architecture, distance metric, and resulting peer selection algorithm. As result of this study, we propose a novel and generic taxonomy for locality algorithms in peer-to-peer networks, aiming to enable a better and more coherent evaluation of any individual locality algorithm.
... After considering the case of M = S 2 , the more general case of M = S G, for G ≤ S, is considered. Figure 5 shows the unfolded connections for LDI (12,4) which contains M=12 nodes. The node degree for this case is S =4 and the value of G is 3. ...
... In fact, because there are fewer groups than nodes within a group, then some nodes can be redundantly reached from other groups resulting in multiple paths between some sources and destinations. For example, two paths from node 5 to node 9 are shown in bold in Figure 5. (12,4) showing multiple paths from node 5 to node 9. ...
... However, in addition to their applicability to routing random traffic in circuit switched networks, the LDI graphs may be applied whenever low diameter directed graphs are needed. Examples of such applications are overlay networks [4], WDM routing in light-wave networks [13] and the pre-scheduling of collective communication patterns [9,10]. It is worth noting that routing in LDI is performed using modular arithmetic operations, which leads to efficient hardware implementations. ...
A new class of low diameter interconnections (LDI) is proposed for high-performance computer systems that are augmented with circuit switching networks. In these systems, the network is configured to match the communication patterns of applications, when these patterns exhibit temporal locality, and to embed a logical topology to route traffic that does not exhibit locality. The new LDI topology is a surprisingly simple directed graph which minimizes the network diameter for a given node degree and number of nodes. It can be easily embedded in circuit switching networks to route random traffic with high bandwidth and low latency
... This causes additional concerns on the cost of maintaining such a network. Adaptive approaches such as Omicron [6] can deal better with this phenomenon. However, there are no studies that capture the behavior of users in very large scale environments. ...
... In addition, an incrementally expandable algorithm has been designed to adapt the exponentially growing de Bruijn graphs to the incrementally expandable P2P systems. Detailed description of the Omicron mechanisms and algorithms are provided in [6]. ...
Collaborative virtual environments are distributed systems that offer collaboration environments containing multiple types of media sources. Taking into account additional constraints such as bounded latency scalability becomes a challenging problem in such systems. This work explores the design space of developing scalable, heterogeneous and dynamic virtual environments based on distributed hash tables. Such structures have been exploited for large scale peer-to-peer systems to provide fast lookup operations in a well distributed way. In This work it is argued that large scale collaborative virtual environments can operate more effectively using peer-to-peer communication paradigms. Such approaches reduce the high communication cost and avoid potential performance bottlenecks of centralized approaches. The load-balancing issues are addressed by distributing the maintenance responsibility for storing and providing dynamic, persistent information that describes the state of the virtual environment. Heterogeneity in participants' capabilities is also being considered.
... In this regard, grid systems interconnect storage systems, computer clusters, instruments, and existing share resources, such as data, storage, software applications, equipment, and CPU time. Over several decades P2P networks have become an effective way for distributed resources for communication and cooperation among nodes [27][28][29]. One of the most popular services proposed by P2P is file sharing (e.g., Gnutella). ...
Resource discovery on different unstructured and dynamic networks such as grid, peer-to-peer, and cloud networks is an inevitable challenging issue. The primary method for resource discovery on the unstructured networks is flooding a query on the network. All existing flooding algorithms for unstructured networks generate almost high additional duplicated queries. This high duplication of the unstructured networks causes a lot of network traffic. This paper, therefore, proposes a novel flexible Distributed Dynamic backbone-based Flooding (DDBF) algorithm for distributed unstructured networks. This paper explores Grid middleware, Peer-to-Peer (P2P) paradigm, and cloud networks resource discovery requirements and it proposes flooding algorithm based on the P2P networks using simulation. To evaluate and prove DDBF algorithm we, first, evaluated it on four fixed network topologies along with two different query flooder distributions, then we evaluated it with one dynamic network topology. The performance of the proposed DDBF algorithm was assessed with five different metrics. The result showed a dramatic decrease in the number of engaged flooder nodes, the number of duplicated queries and consequently, network delay compared with the state-of-the-art algorithms.
... With the mature of heuristic algorithm, many researchers proposed MA data fusion techniques based on genetic algorithm. However, the generic algorithm fusion energy is the function of time which can't reduce the data transmission, so the energy can't be saved and it has lots of shortcomings [5]. This paper proposes a correcting fusion algorithm for sensor network data communication. ...
... Darlagiannis etal. recently presented Omicron [3], a design study for structured overlay networks. The idea is to split the algorithm that each overlay member traditionally executes into a number of simpler services that have different requirements. ...
Implementing overlay software is non-trivial. Current projects commonly build overlays or intermediate frameworks on top of network-ing abstractions. However, a number of frameworks were recently pre-sented that aim to provide support at higher abstraction levels. We ex-plore the design space of such abstractions and show that some levels are still ignored by major frameworks, especially when it comes to high level support for topology adaptation and overlay quality of service (QoS). We present a novel approach to overlay design that aims to integrate dif-ferent overlays and makes them available for topology adaption and se-lection. The approach, named Node Views, heavily borrows from achieve-ments in active databases and distributed view maintenance. It decou-ples generic components from specific overlay implementations and thus facilitates the pluggable development of overlay software.
... Due to constraints on bandwidth, memory, and processing speed, nodes in a network have very different capacities. However, most attempts start with a naturally homogeneous system and then graft in heterogeneity [8, 10]. While some systems [5, 13, 15] (typically unstructured) are by nature heterogeneous, they neither anticipate nor rigorously analyze the benefit. ...
Exhaustive search in large-scale peer-to-peer systems is com-plicated by heterogeneity, crashes, node churn, hotspots, and weakly structured data. While existing approaches solve some of these problems, the BubbleStorm system offers a naturally integrated and simple solution based on random multigraphs. The simplicity of this mathematical structure allows us to rigorously analyze even the heterogeneous case. We present a new communication primitive, named bubble-cast, which induces subgraphs (bubbles) of controlled size. It can incrementally enlarge a bubble, but operates in par-allel. Further, the underlying topology deals easily with crashes and node churn as maintenance operations are lo-cal, atomic, and minimally disruptive. In spite of this, it preserves the global random structure of a simple permuta-tion. When combined into the BubbleStorm system, bub-blecast on this topology performs exhaustive search with adjustable probabilistic guarantees and no hotspots. If ev-ery query must rendezvous with every datum, this approach has optimal per-node bandwidth complexity. Indeed, rather than suffering from heterogeneity, it is exploited fully.
... Therefore , if communication quality affects market share, when competition among VoIP companies intensifies, P2P VoIP platform providers should propose a reward mechanism to solve the problem of non-cooperative supernodes in communication networks. Regarding the implementation of a reward mechanism, although most existing popular commercial P2P VoIP networks have not adopted reward schemes for improving the QoS, several studies have investigated incentive mechanisms in P2P routing252627. In addition, a few micro-payment mechanisms for P2P networks [28] have been proposed in recent years. ...
As we have seen, P2P VoIP software, such as Skype, has emerged from the current generation of telecommunication systems. However,
establishing communication applications based on P2P networks without considering the operational models of the Internet presents
potential dangers. In this study, we treat a VoIP telephone conversation as a dynamic game and compare the ex post reward
mechanism with the ex ante reward mechanism in forward versus discard and QoS routing. Our simulation results point out that
the ex post reward mechanism is better than the ex ante reward mechanism in forward versus discard; however, the opposite
holds true in QoS routing when the expected number of periods is sufficiently large. In addition, this study considers the
reward mechanisms and the platform provider’s benefit to find the optimal number of supernodes in a transmission path.
KeywordsReward mechanisms-Moral hazard-Communication-P2P VoIP networks
... Studies related to hierarchical overlay networks try to improve the canonical overlay networks. When a hierarchical architecture is considered, it is necessary to take into account the different trade-offs that arise with these types of architectures [29]. It is demonstrated that these architectures have benefits [30], [31] in comparison with the canonical counterparts. ...
The IETF P2PSIP WG is currently standardising a protocol for distributed multimedia services combining the media session functionality of SIP and the decentralised distribution and localisation of resources in peer-to-peer networks. The current P2PSIP scenarios only consider the infrastructure for the connectivity inside a single domain. This paper proposes an extension of the current work to a hierarchical multi-domain scenario: a two level hierarchical peer-to-peer overlay architecture for the interconnection of different P2PSIP domains. The purpose is the creation of a global decentralised multimedia services in enterprises, ISPs or community networks. We present a study of the routing performance and routing state in the particular case of a two-level distributed hash table hierarchy that uses Kademlia. The study is supported by an analytical model and its validation by a peer-to-peer simulator.
... Since the advent of these systems a number of proposals using for instance Supernodes [54], [55] or decentralized structured approaches based on hash search and indexing565758 have been developed. Further, some systems use hybrid approaches that combine client/server aspects with P2P structured and unstructured concepts616263 . Structured P2P networks index content using hash indices. ...
Due to the technical developments in electronics the amount of digital content is continuously increasing. In order to make digital content respectively multimedia content available to potentially large and geographically distributed consumer populations, Content Distribution Networks (CDNs) are used. The main task of current CDNs is the efficient delivery and increased availability of content to the consumer. This area has been subject to research for several years. Modern CDN solutions aim to additionally automate the CDN management. Furthermore, modern applications do not just perform retrieval or access operations on content, but also create content, modify content, actively place content at appropriate locations of the infrastructure, etc. If these operations are also supported by the distribution infrastructure, we call the infrastructure Content Networks (CN) instead of CDN. In order to solve the major challenges of future CNs, researchers from different communities have to collaborate, based on a common terminology. It is the aim of this paper, to contribute to such a terminology, to summarize the state-of-the-art, and to highlight and discuss some grand challenges for CNs that we have identified. Our conception of these challenges is supported by the answers to a questionnaire we received from many leading European research groups in the field.
... P2P overlay networks have been studied extensively for different application scenarios such as file-sharing, distributed computing, collaboration, and video streaming. Most of this work is focused on network control and search, while data transport follows an end-to-end paradigm [5] . Recently some work addressed data transport and routing aspects using P2P overlays, e.g., [1], [2], and [6]. ...
Multi-hop ad hoc networks are envisioned to be an integral part of future mobile networks. Today, however, ad hoc networks lack the necessary resilience and reliability. This work proposes to use P2P routing overlays to enhance the dependability and performance of multi-hop ad hoc networks for devices that are subscribed to a provider.
... So kann es sinnvoll sein, Untergruppen im Overlay zu bilden, sei es aus Gründen der Effizienz (z.B. Multicast oder Hierarchien), der Sicherheit (Gruppenverschlüsselung oder gezielter Einsatz von TLS) oder der Fehlertoleranz (Clustering [DMS04] ...
... Building on the large body of literature on query modification and optimisation, we can imagine a number of ways to investigate for pre-optimising these statements. This is most interesting for views of views and for merging view definitions when sending them over the wire (like in gossip overlays [16] or hierarchical environments [17]). This paper presented Node Views, a novel approach to overlay design frameworks that enables support for topology rules, maintenance, adaptation and selection at a very high level. ...
Implementing overlay software is non-trivial. Current projects build overlays or intermediate frameworks on top of low-level networking abstractions. This leaves implementing the topologies, their maintenance and optimisation strategies, and the routing to the developer.
We take a novel approach to overlay implementation by modelling topologies as a distributed database. This approach, named “Node Views”, abstracts from low-level issues like I/O and message handling. Instead, it moves ranking nodes and selecting neighbours into the heart of the overlay software development process. It decouples maintenance components in overlay software and allows implementing them in a generic, configurable way for pluggable integration in frameworks.
... Bloom filters [2] are used in Oceanstore [13]) are selected to improve the accuracy or the efficiency respectively of the P2P systems. Finally, approaches such as Omicron [4] follow hybrid solutions to achieve higher adaptation on the dynamically changing factors of the environment. ...
... Second, including not only the cost of JXTA in the communication times, but also in the management of the peers. This will be done in collaboration with the University of Darmstadt [8]. Some new features have also to be studied for JNGI. ...
The aim of P2P computing is to build virtual computing systems dedicated to large-scale computational problems. JXTA (JuXTApose) proposes an underlying infrastructure on which JNGI (Jerome, Neelakanth, Greg, and Ilya: first names of the creators), one of the first P2P decentralized computing frameworks is built. In order to test this framework, we have built a tool named P2PPerf which allows us to study the behavior of JNGI and to optimize it according to our simulation results.
... For this purpose, a decentralized algorithm to split and merge the clusters is described in. 22 Moreover, a similar random walk mechanism may be applied when peers are becoming reliable enough to be assigned more critical roles (i.e., Indexers and Maintainers). ...
Several types of Content Distribution Networks are being deployed over the Internet today, based on different architectures to meet their requirements (e.g., scalability, efficiency and resiliency). Peer-to-peer (P2P) based Content Distribution Networks are promising approaches that have several advantages. Structured P2P networks, for instance, take a proactive approach and provide efficient routing mechanisms. Nevertheless, their maintenance can increase considerably in highly dynamic P2P environments. In order to address this issue, a two-tier architecture called Omicron that combines a structured overlay network with a clustering mechanism is suggested in a hybrid scheme.
In this paper, we examine several sampling algorithms utilized in the aforementioned hybrid network that collect local information in order to apply a selective join procedure. Additionally, we apply the sampling algorithms on Chord in order to evaluate sampling as a general information gathering mechanism. The algorithms are based mostly on random walks inside the overlay networks. The aim of the selective join procedure is to provide a well balanced and stable overlay infrastructure that can easily overcome the unreliable behavior of the autonomous peers that constitute the network. The sampling algorithms are evaluated using simulation experiments as well as probabilistic analysis where several properties related to the graph structure are revealed.
... A number of papers have been dealing with hierarchical overlay networks to improve the performance of canonical overlay networks. Thus, it is necessary to take into account the different trade-offs that arise with these types of architecture and what has to be addressed [4] when adopting a hierarchical approach. Nevertheless, the benefits of hierarchical architectures have been widely studied [7], [8]. ...
The scalability properties of DHT based overlay networks is considered satisfactory. However, in large scale systems this might still cause a problem since they have a logarithmic complexity depending. Further, they only provide a one dimensional structure and do not make use on inherent clustering properties of some applications (e.g. P2PVoIP or locality aware overlays). Thus, structures based on a hierarchical approach can have performance as well as structural advantages. In this paper, a generic hierarchical architecture based on super-peers is presented where a peer ID is composed by a prefix ID and a suffix ID. Prefix ID is only routed at the super-peer level and the Suffix ID at the peer level. We specifically analyse the Routing Performance of this approach within the context of two specific overlays, viz. CAN and Kademlia.
In this study, a novel overlay architecture for constructing hierarchical and scalable clustering of Peer-to-Peer (P2P) networks is proposed. The proposed architecture attempts to enhance the clustering of peers by incorporating join, split, merge and cluster leader election mechanisms in a fully distributed manner. It takes delay proximity of peers into account as distance measure. By constructing hierarchical clustering of peers, the control message overhead and maintenance such as host departure/host join overhead are decreased. Theoretical comparisons on overheads of the proposed system with that of other systems from literature are studied. The control mechanism for dynamic peer behavior of the architecture is tested over PlanetLab. The performance metrics used are end-to-end delay, diameter, cluster head distance, occupancy rate, peer join latency, accuracy and correctness. The test results are compared with Hierarchical Ring Tree (HRT) and mOverlay architecture. In addition, a P2P video streaming application is run over the proposed network overlay. Streaming tests show that video streaming applications perform well in terms of received video quality if hierarchical clusters considering delay proximity are used as underlying network architecture.
The peripheral malfunction will cause the nodes fail to work when there is no central control node in the large scale sensor network. The network cant guarantee the continuity of the communication and service and the communication efficiency will be low in this situation. In order to fix the issue, the paper proposes a correcting fusion algorithm for the sensor network data communication. When the collision of the communication data happens in the sensor network due to the failure of the nodes, the correcting fusion algorithm can ensure the effective data fusion between the fault nodes and other nodes to guarantee the communication to the maximum extent. The experiment indicates the algorithm can increase the efficiency of the sensor network to achieve the satisfactory results.
In this paper we analyze the functional and non-functional requirements of peer-to-peer (P2P) systems that go beyond the needs of the well explored file-sharing P2P systems. Four basic subcategories are suggested to classify the non-functional requirements: adaptability, efficiency, validity and trust. Similarly, the functional requirements are divided into user-triggered and system-triggered functions. Then, we present and discuss several existing solutions following different design approaches in order to reveal their suitability for different types of P2P applications. More particularly, we capture the characteristics of structured and unstructured overlay networks. For the structured solutions, we investigate distributed hash tables and lexicographic graph based solutions. For the unstructured approaches, we discuss both hierarchical and non-hierarchical overlay networks and emphasize on small-world and power-law networks. Finally, we provide a set of demanding application classes such as voice over IP and massive multiplayer online games that may be benefited by utilizing the P2P technology.
Structured overlay networks for Peer-to-Peer systems (e.g. based on Distributed Hash Tables) use proactive mechanisms to provide
efficient indexing functionality for advertised resources. The majority of their occurrences in proposed systems (e.g. Chord,
Pastry) provide upper bounds (logarithmic complexity with respect to the size of the graph representing the network) on the
communication cost in worst case scenarios and their performance is superior compared to unstructured alternatives. However,
in particular (empirically observed) scenarios where the popularity of the advertised resources follows a distribution considerably
different from the uniform distribution, structured P2P networks may perform inferiorly compared to well designed unstructured
P2P networks that exploit effectively the resource popularity distribution. In order to address this issue, a very simple
caching mechanism is suggested in this paper that preserves the theoretical superiority of structured overlay networks regardless
of the popularity of the advertised resources. Moreover, the churn effect observed in Peer-to-Peer systems is considered.
The proposed mechanism is evaluated using simulation experimesnts.
The intrinsic properties of the employed graphs in designing peer-to-peer overlay networks are crucial for the performance of the deployed peer-to-peer systems. Several structured topologies have been proposed based on meshes, enhanced rings, redundant tree structures, etc. Among them, de Bruijn graphs are promising alternatives since they provide some crucial asymptotically optimal characteristics. In this paper, we discuss the necessary algorithms and protocol messages to develop efficiently the employed routing procedure of Omicron, which is a hybrid overlay network based on de Bruijn graphs enriched with clustering and role specialization mechanisms. Enhancements of the original de Bruijn structure are advised to cope with the intrinsic issue of uneven distribution of the routing workload. The developed system is evaluated and compared with Chord, which is used as the reference point. The superiority of de Bruijn based overlay networks with respect to scalability is quantitatively demonstrated using simulation experiments. Further, the ability of the two systems to exploit the underlying network is investigated
Today, peer-to-peer (P2P) networks, e.g., lesharing net- works like Gnutella, are specialised towards specic purposes. New P2P networks have to be created and installed to support new purposes. Fur- thermore, current P2P networks are almost impossible to use in business processes, because their lack of control makes them too unreliable. This paper addresses these problems by introducing a new middleware for P2P networks. The middleware supports the deployment and use of services inside a P2P network, independent of their purpose. It follows a modular design, thus allowing adaption through plug-ins. It includes negotiation mechanisms which can be used to congure a service according to a specic need. Furthermore, the middleware makes use of strong identi- ties and oers mechanisms enabling the management of services through market forces, thus allowing the creation of legally enforceable contracts. Thus, it allows to reliably compose services into new value-added ser- vices. This paper also briey describes the middleware's implementation and application in lesharing and wireless LAN scenarios.
As an infrastructure for data distribution, overlay networks have to feature efficient routing and adequate robustness to achieve fast and accurate data distribution in the environment with node churn. Considering that the existing overlay networks mostly focus on single optimization objective and fail to ensure routing efficiency and robustness simultaneously, a hybrid overlay network for content-based data distribution - Anadem is proposed in this paper. Anadem achieves a better compromise between routing efficiency and robustness by combining the inter-cluster multiple structured topologies with the intra-cluster unstructured topologies. Anadem also provides mechanisms for dynamic concurrent cluster creation, cluster departure and load balance to make data distribution more adaptive to the dynamic network environment. Experimental results reveal that compared with existing overlay networks, Anadem can support fast and accurate content-based data distribution even when large amount of nodes fail in the system.
Peer-to-Peer (P2P) technology has been widely applied in today's Internet. Unstructured P2P systems, where peers connect with each other to form dynamic, flexible and scalable networks, are commonly used for resource sharing. Resource location in unstructured P2P systems tends to use "blind search" strategies (i.e. flooding and random walk) for its simplicity and low maintenance cost. However, due to lack of location information, the performance of "blind search" is heavily relied on the network topology. Topology adaptation, by properly adjusting the topology of the P2P overlay network, is a promising approach to improve the search performance. In this paper, we use Relative Search Betweenness (RSB) to estimate nodes' search ability. A RSB-based topology adaptation algorithm (RSB-Topo) is proposed, where peers spontaneously adjust their connections to achieve better performance. Simulation results show that our algorithm could greatly increase search success rate and search coverage, and also decrease response delay to improve the search performance in unstructured P2P networks.
Today's peer-to-peer applications benefit from the fact that many users offer their resources (mostly inform of files). Those resources are mainly connected via relatively low-bandwidth, asymmetric access networks (such as ADSL or cable modems), which make it hard to realize the streaming of video data. Thus, audio visual content is usually downloaded and not streamed in today's peer-to-peer (P2P) systems. In order to provide streaming support it is necessary to take into account the asymmetric character of the up-load and download links. We show that by making use of multiple description coded (MDC) video and the fact that single descriptions can be sent from different peers, streaming in peer-to-peer applications is feasible. The paper discusses the different issues related to this topic. It explains MDC and compares it to hierarchically layered encoded video (HLEV). Further, the conditions under which MDC can be used for P2P streaming are discussed and it is shown how it can be deployed in a P2P environment.
This paper studies the behavior of the diameter and the average distance between vertices of the line digraph of a given digraph. The results obtained are then applied to the so-called (d, k) digraph problem, that is, to maximize the number of vertices in a digraph of maximum out-degree d and diameter k. By line digraph iterations it is possible to construct digraphs with a number of vertices larger than (d<sup>2</sup>- l)/d<sup>2</sup>times the (nonattainable) Moore bound. In particular, this solves the (d, k) digraph problem for k = 2. Also, the line digraph technique provides us with a simple local routing algorithm for the corresponding networks.
The Project JXTA protocols establish a virtual network overlay on top of the Internet, allowing peers to directly interact and self-organize independently of their network connectivity and domain topology (firewalls or NATs). Project JXTA enables application developers, not just network administrators to design network topology that best match their application requirements. Multiple ad hoc virtual networks can be created and dynamically mapped into one physical network unleashing a richer multi-dimensional virtual network world. Project JXTA is looking ahead where billion of network services, all addressable on the network will be able to discover and interact with each other in an ad doc and decentralized manner through the formation of a multitude of virtual networks.
A fundamental theoretical challenge in peer-to-peer systems is proving statements about the evolution of the system while nodes are continuously joining and leaving. Because the system will operate for an infinite time, performance measures based on runtime are uninformative; instead, we must study the rate at which nodes consume resources in order to maintain the system state.
This “maintenance bandwidth” depends on the rate at which nodes tend to enter and leave the system. In this paper, we formalize this dependence. Having done so, we analyze the Chord peer-to-peer protocol. We show that Chord’s maintenance bandwidth to handle concurrent node arrivals and departures is near optimal, exceeding the lower bound by only a logarithmic factor. We also outline and analyze an algorithm that converges to a correct routing state from an arbitrary initial condition.
We propose a new approach for constructing P2P networks based on a dynamic decomposition of a continuous space into cells corresponding to processors. We demonstrate the power of these design rules by suggesting two new architectures, one for DHT (Distributed Hash Table) and the other for dynamic expander networks. The DHT network, which we call Distance Halving allows logarithmic routing and load, while preserving constant degrees. It offers an optimal tradeoff between the degree and the dilation in the sense that degree d guarantees a dilation of O(log dn). Another advantage over previous constructions is its relative simplicity. A major new contribution of this construction is a dynamic caching technique that maintains low load and storage even under the occurrence of hot spots. Our second construction builds a network that is guaranteed to be an expander. The resulting topologies are simple to maintain and implement. Their simplicity makes it easy to modify and add protocols. A small variation yields a DHT which is robust against random faults. Finally we show that, using our approach, it is possible to construct any family of constant degree graphs in a dynamic environment, though with worst parameters. Therefore we expect that more distributed data structures could be designed and implemented in a dynamic environment.
We describe a peer-to-peer system which has provable consistency and performance in a fault-prone environment. Our system routes queries and locates nodes using a novel XOR-based metric topology that simplifies the algorithm and facilitates our proof. The topology has the property that every message exchanged conveys or reinforces useful contact information. The system exploits this information to send parallel, asynchronous query messages that tolerate node failures without imposing timeout delays on users.
Virtual overlay networks, such as virtual private networks or peer-to-peer services, can be seen as a new paradigm for providing multi-service networks. Virtual overlay networks may offer customized services to a specified community while providing a high degree of flexibility in the usage of shared resources. This paper examines the requirements of operating dynamic overlays, in particular, for peer-to-peer services. The analysis is based on extensive measurement studies performed on the global Gnutella network during operation. The obtained results indicate limitations in the scalability of native p2p overlays, suggesting the need of a control scheme for efficiency reasons. As an enabling infrastructure to implement a distributed control scheme for p2p overlays a so-called Application-Layer Active Networking (ALAN) platform has been chosen. Based on Application-Layer Active Networking, Active Virtual Peers (AVP) are introduced as the main concept for dynamic operation and management of peer-to-peer overlay networks. AVPs facilitate policy enforcement or performance management by means of self-organization, predominantly on the application layer with minimum interference on lower layers.
Consider an arbitrary distributed network in which large numbers of objects are continuously being created, replicated, and destroyed. A basic problem arising in such an environment is that of organizing a distributed directory service for locating object copies. In this paper, we present a new {\em data tracking scheme}\/ for locating nearby copies of objects in arbitrary distributed environments. Our tracking scheme supports efficient accesses to data objects while keeping the local memory overhead low. In particular, our tracking scheme achieves an expected $\polylog(n)$-approximation in the cost of any access operation, for an arbitrary network. The memory overhead incurred by our scheme is $O(\polylog(n))$ times the maximum number of objects stored at any node, with high probability. We also show that our tracking scheme adapts well to dynamic changes in the network.
We consider the problems of routing and sorting on a de Bruijn
network. First, we show that any deterministic oblivious routing scheme
for permutation routing on a d-ary de Bruijn network with
N=d<sup>n</sup> nodes, in the worst case, will take Ω(√N)
steps under the single-port model. This improves the existing lower
bounds provided d is not a constant. We also show that the lower bound
is indeed a tight one. Second, we present a deterministic nonoblivious
permutation routing algorithm which runs in O(d.n<sup>2</sup>) time on a
d-ary de Bruijn network with N=d<sup>n</sup> nodes. This algorithm is
currently the fastest known nonoblivious deterministic routing algorithm
for de Bruijn networks of arbitrary degree. Finally, we present an
efficient general sorting algorithm for the de Bruijn networks of
arbitrary degree. This algorithm is the best sorting algorithm known so
far. It runs in O((log d).d.n<sup>2</sup>) time for directed de Bruijn
network with d<sup>n</sup> nodes, degree d, and diameter n. As a
corollary, we show that on a binary de Bruijn network of Nnodes, our
sorting scheme requires at most 2 log<sup>2</sup> Nsteps
De Bruijn graphs, both directed and undirected, have received
considerable attention as architecture for interconnection networks. In
this paper, we focus on undirected de Bruijn networks of radix d and
dimension 0, denoted by UB(d, 0). We first discuss the shortest-path
routing problem. We present properties of the shortest paths between any
two vertices of UB(d, 0) and propose two shortest-path routing
algorithms, one of which has linear time complexity. Secondly, we study
the transmitting problem. We establish a lower bound for the optimal
transmitting time which implies in particular that the optimal
transmitting problem is trivial for UB(d, 0) when d⩾5. We present a
transmitting scheme on undirected binary de Bruijn networks UB(2, n)
with transmitting time n-1 for n⩾5, and conjecture that the optimal
transmitting time is n-1 for UB(2, n), and n for U8(3, n) and UB(4, n)
Gnutella is a classical Peer-to-Peer network designed for file-sharing. The absence of pure servers is one of its main properties, given that every Gnutella host is client and host in one. It uses the resources of the participants to distribute content, e.g. mp3 compressed audio files, and shares the processing capacity to provide the routing and searching capabilities for the network.
A content-addressable network (CAN) is a distributed lookup table that can be used to implement peer-to-peer (P2P) systems. A CAN allows the discovery and location of data and/or resources, identi ed by keys, in a distributed network (e.g., Internet), in absence of centralized server or any hierarchical organization. Several networks have been recently described in the literature, and some of them have led to the development of experimental systems. We present a new CAN, called d2b. Its main characteristics are: simplicity, provability, and scalability. d2b allows the number of nodes n to vary between 1 and jKj where K is the set of keys managed by the network. In term of performances, any join or leave of a user implies a constant expected number of link modi cations, and, with high probability (w.h.p.), at most O(log n) link modi cations.
Necklaces and scalability of Kautz digraphs Vasilios Darlagiannis received his Dipl-Inf degree from the Technical University of Crete, Greece. He holds an MASc in Electrical and
- Pavel Tvrdik
Pavel Tvrdik, Necklaces and scalability of Kautz digraphs, in Sixth IEEE Symposium on Parallel and Distributed Processing, October 1994. Vasilios Darlagiannis received his Dipl-Inf degree from the Technical University of Crete, Greece. He holds an MASc in Electrical and Computer Engineering obtained from the University of