Overlay Partition: Iterative Detection and Proactive Recovery

Size: px
Start display at page:

Download "Overlay Partition: Iterative Detection and Proactive Recovery"

Transcription

1 Overlay Partition: Iterative Detection and Proactive Recovery Tongqing Qiu, Edward Chan and Guihai Chen State Key Laboratory for Novel Software Technology, Nanjing University, China Department of Computer Science, City University of Hong Kong, China Abstract Overlay networks provide infrastructures for a large variety of Internet applications, like file sharing, online gaming, and multimedia streaming. However, these networks often face unexpected node failures and network disconnections, causing the overlay to be partitioned into several components, which may seriously affect the performance of the network. In this paper, we analyze the cause of overplay partitions and its impact on the efficiency of the system. After explaining the notion of half-life, a measurement used to describe the evolution of peers in overlay network, we propose a new construct called halfsuccess as a measure of critical partition. Furthermore, we propose an iterative method for the detection of potential partitions and a proactive strategy for the prevention of such partitions. Simulation experiments show that our detection method can efficiently find almost all of the cut vertices at a low cost. In addition, we demonstrate that the proposed proactive scheme is more effective and much faster than reactive approaches. I. INTRODUCTION In the last few years, overlay networks have rapidly evolved into a promising platform for deploying new applications and services in the Internet [1]. An overlay network is a computer network which is built on the top of another network. Nodes in the overlay can be thought as being connected by virtual or logical links, each of which corresponds to a path, perhaps through many physical links, in the underlying network. Overlay networks can be constructed in order to route messages to destinations not specified by an IP address in peerto-peer systems [2], [3], [4], [5]. However, application-level overlay usually face unpredictable failures. This is because each node/peer is viewed as an independent entity, which can join and leave the network at will. Moreover, the underlying network is not always reliable, since packets can be lost due to transient problems, network link failures and in some cases malicious attacks. In order to provide quality services, an overlay network needs to guarantee reliability and robustness. Two basic elements to achieve this are failure detection and recovery. Failure detection algorithms can be broadly classified as either active or passive [6]. In the active approach, a node periodically sends keep-live messages; while a passive one only uses data packets to convey liveness information. While both methods can discover failures in specific nodes or links, they may not be able to detect cumulative effect of failures at the system level. Regarding failure recovery, most recent studies focus on a reactive approach to the resilience of overlay i.e. how to recover after failure is detected. The major drawback of the reactive approach is that additional information needs to be collected after a failure has been detected, resulting in a significant delay in reconstructing the overlay. More importantly, it is not always possible to collect enough information for the recovery process after the failure has occurred, especially when the overlay has been partitioned due to the failures. In this paper, we analyze a serious failure condition, overlay partition, and propose suitable detection and recovery mechanisms. Our main contributions are as follows: To the best of our knowledge, this is the first comprehensive study of overlay partitions that includes both theoretical analysis and experimental evaluation. We used the notion of half-life and propose a sign of critical partition: half-success. We formulate an iterative method to detect potential partitions in the network. Our method is distributed, lightweight and is faster than similar algorithms. We also propose a proactive recovery scheme which can recover from node departures much faster than reactive methods. II. PRELIMINARIES We consider an application overlay as an undirected connected graph G =(V,E), where V is a set of nodes and E is the set of logical links. An edge xy E means that x knows a direct way to send message to y. G is called k-connected when no two vertices of G are separated by fewer than k other vertices. The greatest integer k such that G is k-connected is the connectivity κ(g) of G. Recall that cut vertex is a vertex whose removal would disconnect the connected component containing it, and bridge means a edge whose removal would lead to this kind of disconnection. A biconnected graph is a graph with no cut vertices. A biconnected component (bcc) is a maximal biconnected subgraph. In the following two sections, we will illustrate that overlay partition occurs frequently and has a major impact on the efficiency of the system, and that it is necessary to effectively detect overlay partition and quickly recover from partition. A. The Causes of Overlay Partition In order to find the causes of overlay partition, we first address the following questions: How many node failures will /07/$ IEEE

2 result in the partition of an overlay network and how long does it take? More generally, an overlay graph G with connectivity κ(g) will be disconnected when at least κ(g) nodes fail at the same time. We call these κ(g) nodes critical nodes. The remaining ones are normal. Overlay networks are usually sparse: in overlay multicast systems or unstructured P2P systems, the minimum degree δ(g) is usually a small integer; even in structured P2P systems, δ(g) is usually no larger than O(log N). Given that κ(g) δ(g) and κ(g) << N, the number of critical nodes is much smaller than the normal ones. So we focus our study on partitions among the normal nodes. Consider a model of arrivals and departure [7]: nodes arrive according to a Poisson process with rate λ and depart according to an exponential distribution with rate µ (i.e. the expected lifetime is 1/µ). Given that only removal leads to partition, the probability p that a node fails in time τ is 1 e µτ. We call the interval from the initial time when G is connected to the time when the normal nodes are partitioned partition time τ 0. If the failed nodes are just those critical nodes, the overlay will be partitioned after κ(g) nodes fail. So the lower bound of the time to be disconnected is shown in equation 1, by letting κ(g)/n =1 e µτ, τmin 0 = 1 µ ln N (1) N κ(g) Under another extreme condition where the first failed N κ(g) are normal nodes, we can get the corresponding upper bound: τmax 0 = 1 µ ln N (2) κ(g) In reality, the remaining N κ(g) nodes are not always normal nodes. Moreover, when different nodes fail continually, normal nodes may turn into critical nodes. So the upper bound of partition time is quite loose. Based on the range τ 0 1 µ [ln N N κ(g), ln N κ(g) ], we can obtain additional information. First of all, the partition time τ 0 is influenced by two parameters: the activity µ and the connectivity κ(g). More specifically, the change of time is intuitively proportional to node s life time 1/µ. Second, if only the analytical range is meaningful (τmin 0 τ max), 0 no matter how κ(g) changes, in other words, regardless of what the overlay is like, the partition time can just be τt 0 =(1/µ)ln2, when κ(g) =N/2. Figure 1 shows the existence of this value. It is worth mentioning that Liben-Nowell et al. [8] use half-life, the time required for half of nodes to depart, to measure the evolution of P2P systems. Their notion of half-life is just τt 0. In their paper, they did not explain the reason for using half-life. We give one reason here: Half-life means how long the overlay will be partitioned regardless of the topology of the overlay. Third, because κ(g) is much smaller than N, the partition time is rather short. The trace from PlanetLab [9] also illustrates that partition occurs frequently. Moreover, due to The range of partition time Half-life Upper bound Lower bound The connectivity k(g) Fig. 1. Half life. We choose µ =1and N = 100 the distributed, even random organization of overlay, cases of κ(g) =1and τ 0 0 are very common. In this condition, the failure of one node can lead to the disconnection of the whole overlay. Consequently, one possible approach to avoid potential partitions is to periodically exploit cut vertices. It is obvious that after the first partition, the connectivity of deduced graph κ(g ) may be larger, smaller or equal to that of original graph. So the time when the next partition occurs (τ 1,τ 2,...) is unexpected, although eventually τ 0. Note that the range of τ 0, is determined by both the overlay topology (connectivity) and the number of failed nodes (activity). While our partition detection method can partially describe the topology, it is impossible to know the precise number of failed nodes in a distributed system. Furthermore, different nodes have various positions and importance in the system. So this factor cannot be directly used as a sign to detect or evaluate partitions. B. The Impact of Partitions We use the successful look-up ratio (SLR) of requests as the basic metric to examine the impact of partition. It is defined as the ratio of successful query requests to the the total number of requests. If there are m different biconnected components, and the ith component C i has N i nodes, then without any recovery, SLR = m N 2 i N 2 100%, (0 <N i <N;0<m N) (3) Obviously, the number of bbc affects the efficiency of lookups. Under an extreme condition where m = N (i.e. each node is one independent component and N i =1), each node can only find contents located in its own site, and SLR =1/N 2. Moreover, if we view SLR as a function of m, considering that m N i = N, we can easily obtain the lower bound of SLR, and the upper bound SLR max (m) =1 SLR min (m) = 1 m 2(m 1) N (4) m(m 1) N 2 (5)

3 The upper bound represents a condition that m independent nodes are disconnected from the system. When m<<n, this kind of partition can be omitted. It indicates that not all partitions are serious. When it comes to lower bound, equal size of bbc leads to the worst case scenario. Specifically, assuming that the overlay is going to be partitioned into two independent components, in the worst case SLR = 50%. We call this a half-success situation. Conversely, we have every reason to suspect that: The overlay network has been partitioned into two equal size components or even more components under the condition of half-success. We will show in the simulation section that SLR=50% is in fact a reasonable indicator of a critical partition. Overlay partitions not only greatly decrease the successful lookup ratio, but also make failure detection and recovery more difficult. First of all, if the detection method is based on periodic probing within a local range (usually the neighborhood), then it may be impossible for internal nodes within an independent component to discover the failure which occurs on the edge of components. Second, once the partition occurs, the nodes do not know how to make connections between independent components. It is therefore imperative that partitions should be detected as early as possible; and if they occur at all, recovery should be prompt. We will now introduce our iterative detection method and proactive recovery scheme which satisfy these requirements. III. APPROACH FOR MANAGING PARTITIONS In this section we present the key elements in our approach for managing partitions in an overlay network. We first describe the basic detection and recovery mechanisms, and then extend them to handle group failures i.e. partition caused by the failure of a group of nodes. A. Iterative Partition Detection There are several challenges in exploiting cut vertices. First, it is hard to collect the global information in a distributed system and even if it is possible, the overhead is relatively high. Most of the distributed algorithms to find cut vertices are based on the construction of depth first search (DFS) trees [10], [11]. For each node n in the tree except the root, check the neighbors to which n s descendants are connected to. If none of the neighbors of n s descendants are n s ancestors, n is a cut vertex. These distributed methods can find all cut vertices in a static network, but the average cost of detection is about O(N 2 ), where N is the number of nodes. Second, since the overlay is dynamic, cut vertices may fail and cause partitions before they are detected. So it is important to detect potential partition quickly. Take a recently proposed method CAM [12] as an example. It uses flooding to detect cut vertices. Although its overhead only O(N), the flooding will not stop until a specific TTL (time-to-live) value has been reached. Moreover, it cannot automatically tune the flooding range and probing times according to different overlay conditions. Fig. 2. Iterative detection We use an iterative method to locate the cut vertices. In this scheme, each node C iteratively detects some nodes and determines whether C is a cut vertex. Each node has a Boolean matrix to describe the connection relation. For any items x, y, if x and y are connected, (x, y) =true in the matrix. Based on our model of overlay, the relation is reflective, transitive and symmetric 1. The matrix also records the state of each node i.e whether it is in the active or inactive state. Initially, only C s neighbors are in the matrix and they are active. At the beginning of each partition detection cycle, node C asks active nodes to feedback their neighbors information. The message contains the C s IP address, a timestamp, and identification. The receivers respond based on the timestamp and identification of the senders. If a node has already received the message for the same requester or it is not current, then the message is simply dropped. Otherwise, the node will send back its neighborhood connection information to node C. Then C makes the necessary changes in the matrix: it first sets all active nodes to the inactive state, then adds new entries for new nodes and sets them to be active. Moreover, it changes the Boolean values based on the feedback messages. After a simple circulation of transitive closure using Warshall s algorithm [13], the information in the matrix is finalized for that cycle and the number and the size of bbc are determined using this information. For instance, in Figure 2(c) we have two components N 1 =2N 2 =5, respectively. A threshold t is used to limit the number of detection cycles, and the detection process will automatically stop when the number of components m =1. Our detection method is lightweight and effective. Assume that the overlay network has N nodes, and c is the average number of connections of a node. In the worst case, the main overhead of detection is the sum of information retrieval in each detection cycle O((c +(c 2 c)+ +(c t c t 1 ))N) = O(c t N), which is much less than the traditional DFS method O(N 2 ) when c is small. Although this overhead is the same as that of CAM, based on the fact that most of nodes are not cut vertices, in most cases our detection operations will stop before t trials. This means that the detection will be completed much earlier and overhead is reduced. This is a very useful feature since it is difficult to select in advance a good value for a constant threshold to limit the probing range. For instance, 1 The symmetric property is true when we assume that overlay is an undirected graph. In some cases, especially in some structured peer-to-peer systems, it is not true. Since detection is related to transitive closure, its correctness has nothing to do with symmetry.

4 in CAM, a large value for TTL will lead to an exponential growth in overhead. In our approach, setting a larger threshold seldom leads to problems as the detection mechanism usually terminates before the threshold is hit. Note that the detection process begins only when node C is stable. We are able to determine when this stage has been reached since many researchers [8], [14] have studied and characterized the behavior of peers in a stable state. By starting the detection process when the network is in a stable state reduces the unnecessary cost of detection and recovery for those peers that enter and exit the network shortly afterward. B. Proactive Recovery The traditional reactive overlay recovery mechanisms can be broadly classified into two categories based on their different ways of collecting information: central retrieval (CR) and local retrieval (LR). In the first approach, a peer tries to contact a set of well-known nodes and retrieve the locations of other nodes, and then re-establish the connections again. This method can be used in a loosely organized system where no specific requirements on connection exist, like Gnutella [2]. The weakness of this method is that well-known nodes with centralized information are required. Failures of such nodes may cause the whole system to crash. In the local retrieval approach, a peer asks the neighbors to collect information of nodes that are far away. This method is widely used in structured peer-to-peer systems [4], [5], [15], since the peer will qualify each candidate. As we discussed above, a major problem with this approach is that the nodes do not have precise information on how to make connections between independent components when overlay partition occurs. In order to solve this problem, we propose a proactive recovery method in which a recovery plan is devised before the partition occurs. For each cut vertex C, it must leave recovery information for their neighbors. The transfer and acknowledgement of such recovery information can be piggybacked on keep-live messages. Once its neighbors detects that node C fails, they know exactly how to avoid partition by adding new connections. Note that only one single message from the neighbors is needed to complete the recovery process. This is much faster than the reactive approach. Ideally there should be as many logical connections as possible to eliminate the partitions. Given that in a dynamic or error-prone distributed environment, it may not be possible to avoid partitions all the time. As a result the choice of component union is very important. Assuming that components C x and C y are combined, the resulting improvement in SLR after the union operation is: SLR = (N x + N y ) 2 N 2 N 2 x + N 2 y N 2 = 2N xn y N 2 (6) The result shows that the union of two components with more nodes will lead to greater improvement. So we propose a larger-one-first scheme to maximize SLR. We will first sort the m components according to their size N 1 N 2... N m, then add new connections C 1 C 2,C 2 C 3,...C m 1 C m.more Fig. 3. Different recovery conditions specifically, when there is a new connection between C x C y, it means that two neighbors of node C, n i and n j, will be connected together, where n i C x and n j C y. If there are more than one candidates in the choice of neighbors, we choose the one with less degree to balance the load. However, it is easy to proof that all these new connections C 1 C 2,C 2 C 3,...C m 1 C m are bridges. So we need to connect components C m C 1 to eliminate all introduced bridges. Specifically, we will add two connections between the two components when m =2. It is very likely that node C does not have enough time to finish probing and pre-computation of partition components. But we can still use the latest partition information to add new conditions and to ensure connectivity. This feature further illustrates that iterative detection is necessary. Figure 3 shows different recovery conditions before and after the accomplishment of the first detection cycle. In (a), since the failed node believes that there are the three components, the three neighbors are reconnected together to form a ring-like structure. In (b), because the failed node thinks that there are two components, so two connections are added. Note that not all overlays allow the addition of arbitrary connections because of the requirements of overlay construction protocols. We propose that a best effort mechanism be used to maximize the successful lookup ratio, since this allows for some flexibility and variation on how the connections are made. IV. PERFORMANCE EVALUATION We use GT-ITM [16] to generate the topology of an Internetlike network. A number of nodes are then selected as peers in the overlay network. Since the aim of the simulation is to show the correctness of our analysis and the effectiveness of our mechanism under different network conditions, we use generated traffic instead of a real trace so that we can easily control the connectivity of network as well as the scale of the system. The default behavior of peers follows the Poisson arrival and exponential departure model. A. Partition Time We start from a stable overlay with 1,000 nodes. The overlay graph is manually generated with different connectivity κ(g) =1, 4, 8, 16. Then nodes depart according to the exponential distribution with rate parameter µ (i.e., expected node lifetime is 1/µ =10, 20, 30, 40). The unit is time slot.

5 Partition time k(g)=1 k(g)=4 k(g)=8 k(g)= Average life time Successful lookup ratio % 20% 30% 40% 50% 60% 70% half-success 1.6e e+006 2e e e+006 Simulation time(ms) Fig. 4. Partition time Fig. 5. Successful look-ups ratio for varying percentage of node failures Each experimental result is based on the average of 20 simulation runs. Figure 4 shows the relationship between partition time, average life time and overlay connectivity. The result is generally consistent with our analysis in Section II. However, the partition time is not strictly proportional to the average life time, especially when the connectivity is large. This is because in our analysis, we consider a discrete condition where overlay transfers from connection to disconnection with several nodes fail but without the change of connectivity. In fact, when the nodes fail continually, connectivity may decrease due to the loss of redundant paths or connections. So the time of partition is earlier than expected. B. Impact of Overlay Partitions In this section we will illustrate two consequences of overlay partition: decreased SLR and failed recovery. We take the well-known P2P system Tapestry, which uses local retrieval recovery, as an example. There are 1000 nodes, with average round-trip-time (RTT) = 2s between any pair of nodes. Each node starts to generate lookup requests at the interval exponentially distributed about the mean time of 10s. If a look-up does not reach the destination in 20s, it is regarded as a failed look-up. Stabilization routine is trigged every 100s on average. At time = 1800s, when the topology is considered stable, we let a subset of nodes (ranging from 0% to 70%) leave the network simultaneously instead of following the exponential departure model, Recall that in our analysis, SLR = 50% is a reliable indicator for the occurrence of overlay partition. Figure 5 shows than when the SLR drops to 50%, the network never recovers from the failure. Due to the paper limitation, we merely present Tapestry as an example. According to our previous evaluation [17], some other overlay management protocols cannot successfully recover from from the condition of half-success. This result is not surprising since none of these recovery schemes focus on overlay partition. C. Effectiveness of Partition Detection We use three different metrics to evaluate the correctness of partition detection: accuracy rate (AR), false negative rate CDF of accuracy rate % Iterative 10 CAM 0 DFS Detection time(ms) Fig. 6. Cumulative distribution of the detection time (FNR), and false positive rate (FPR). It is obvious that the FNR of our cut vertices detection algorithms is zero. In other words, all nodes that are truly cut vertices will be detected. However, many nodes that are normal ones will be identified as cut vertices. This false positive rate is highly dependent on the threshold t. Table I shows the accuracy rate of partition detection. When t =3, the accuracy rate is more than 95%. Detection time is another important metric to evaluate the efficiency of partition detection methods. We compare our iterative method with other two methods: CAM and DFS. Figure 6 depicts the cumulative distribution of the detection time. It is obvious that our method is faster than the other two methods. For DFS, it has to construct a DFS tree first, so most of the time is wasted on the construction, TABLE I ACCURACY RATE OF PARTITION DETECTION t=1 t=2 t=3 t=4 AR 73.3% 83.2% 97.4% 98.2% FPR 33.8% 20.6% 6.6% 3.4% FNR 0.0% 0.0% 0.0% 0.0%

6 although DFS does provide a more accurate result. According to the description in section III, for CAM, each node will flood its request to nodes TTL hops away. Our method use an iterative method to generate an intelligent response based on the result of each fault detection cycle so that the fault detection time is shorter. Given enough time, however, these two methods achieve nearly the same accuracy rate. V. RELATED WORK Researchers are aware that the topology of overlay has a significant impact on resilience. S. Saroiu et al. [18] show that the failure of a small number of high degree nodes can effectively shatter the overlay network, which makes such a network highly vulnerable in the face of well-constructed, targeted attack. Ripeanu and Foster [19] point out that if every node in an overlay network can have a degree above a constant lower bound, the resistance to malicious attacks can be greatly strengthened. However, their work based on node degrees cannot precisely describe the influence of different nodes, nor can they be used independently to optimize overlay topology. More recently, the overlay partition problem has received some attention from researchers. For instance Sit and Morris [20] propose a cross-check approach for partition detection. By asking other nodes to do random queries and comparing their results with its own, a node can verify whether its view of the network is consistent with others. This method can detect whether a node is isolated. But based on our analysis of the impact of partitions, the isolation of a specific node does not necessarily lead to the crash of the system. Regarding partition recovery, a merging algorithm in proposed in SkipNet [21] to recover from partitions. Unfortunately, this recovery scheme is based on a specific structure and organization. As mentioned earlier in the paper, our detection method is better than CAM proposed by Liu et. al. [12]. When it comes to recovery, CAM simply connects random nodes in different components after finding the cut nodes. To our knowledge, no general, well-designed method has been proposed to repair the partitioning of an overlay network. VI. CONCLUSION In this paper, we investigate an important issue that affects the resilience of overlay: overlay partitions. We analyze the cause and impact of overlay partitions and the simulation results support our analysis. Based on the characteristics of overlay partitions, we further propose an iterative detection method to exploit the cut vertices. This method is distributed and lightweight. Moreover, we suggest a proactive recovery mechanism, which is much faster than recent reactive methods. We are now planning to extend our work by using use real traces as well as different types of peer behavior in our experiments to evaluate our partition detection and recovery method. ACKNOWLEDGEMENTS The work is partly supported by China NSF grants ( , ), Jiangsu Provincial NSF grant (BK ) and China 973 projects (2006CB303004). The Conference Participation is partly supported by Nokia Bridging the World Program. REFERENCES [1] D. G. Andersen, H. Balakrishnan, M. F. Kaashoek, and R. Morris., Resilient overlay networks, in Proceedings of SOSP 2001, [2] [3] I. Stoica, R. Morris, D. Karger, M. F. Kaashoek, and H. Balakrishnan, Chord: A scalable peer-to-peer lookup service for internet applications, in Proceedings of the ACM SIGCOMM, [4] A. Rowstron and P. Druschel, Pastry: Scalable, distributed object location and routing for large-scale peer-to-peer systems, in Middleware 2001, [5] B. Y. Zhao, L. Huang, J. Stribling, S. C. Rhea, A. D. Joseph, and J. D. Kubiatowicz, Tapestry: A resilient global-scale overlay for service deployment, IEEE Journal on Selected Areas in Communications, vol. 22, no. 1, pp , Jan [6] S. Q. Zhuang, D. Geels, I. Stoica, and R. H. Katz, On failure detection algorithms in overlay networks, in Proceedings of INFOCOM 2005, [7] G. Pandurangan, P. Raghavan, and E. Upfal, Building low-diameter p2p networks, in Proceedings of IEEE Symposium on Foundations of Computer Science 2001, [8] D. LibenNowell, H. Balakrishnan, and D. Karger, Analysis of the evolution of peer-to-peer systems, in Proceedings of Principles of Distributed Computing(PODC 2002), [9] A. Mislove, A. Post, A. Haeberlen, and P. Druschel, Experiences in building and operating epost, a reliable peer-to-peer application, in Proceedings of Eurosys 2005, [10] W. Hohberg, How to find biconnected components in distributed networks, Journal of Parrallel and Distributed Computing, vol. 9, no. 4, pp , August [11] J. Park, N. Tokura, T. Masuzawa, and K. hagiharra, Efficient distributed algorithms solving problems about the connectivity of network, in Proceedings of Systems and Computers in Japan, [12] X. Liu, L. Xiao, A. Kreling, and Y. Liu, Optimizing overlay topology by reducing cut vertices, in Proceedings of NOSSDAV 2006, [13] T. H. Cormen, C. E. Leisereson, and R. L. Rivest, Introduction to Algorithms. MIT Press, [14] F. Bustamante and Y. Qiao, Friendships that last: Peer lifespan and its role in p2p protocols, in Proceedings of International Workshop on Web Content Caching and Distribution, [15] S. S. Lam and H. Liu, Failure recovery for structured p2p networks: protocol design and performance evaluation, in ACM SIGMETRICS Performance Evaluation Review, [16] E. W. Zegura, K. L. Calvert, and S. Bhattacharjee., How to model an internetwork, in Proceedings of INFOCOM, [17] Z. Liu, R. Yuan, Z. Li, H. Li, and C. Chen, Survive under high churn in structured p2p systems: evaluation and strategy, in Proceedings of ICCS 2006, [18] S. Saroiu, P. K. Gummadi, and S. D. Gribble, Measuring and analyzing the characteristics of napseter and gnutella hosts, Multimedia Systems, vol. 9, no. 1, pp , [19] M. Ripeanu and I. Foster, Peer-to-peer architecture case study: Gnutella network, in Proceedings of IEEE P2P 2001, [20] E. Sit and R. Morris, Security considerations for peer-to-peer distributed hash tables, in Proceedings of IPTPS 02, [21] N. J. Harvey, J. Dunagan, M. B. Jones, S. Saroiu, M. Theimer, and A. Wolman., Skipnet: A scalable overlay network with practical locality properties, in Proceedings of USITS 03, 2003.

Fault Resilience of Structured P2P Systems

Fault Resilience of Structured P2P Systems Fault Resilience of Structured P2P Systems Zhiyu Liu 1, Guihai Chen 1, Chunfeng Yuan 1, Sanglu Lu 1, and Chengzhong Xu 2 1 National Laboratory of Novel Software Technology, Nanjing University, China 2

More information

Survive Under High Churn in Structured P2P Systems: Evaluation and Strategy

Survive Under High Churn in Structured P2P Systems: Evaluation and Strategy Survive Under High Churn in Structured P2P Systems: Evaluation and Strategy Zhiyu Liu, Ruifeng Yuan, Zhenhua Li, Hongxing Li, and Guihai Chen State Key Laboratory of Novel Software Technology, Nanjing

More information

Should we build Gnutella on a structured overlay? We believe

Should we build Gnutella on a structured overlay? We believe Should we build on a structured overlay? Miguel Castro, Manuel Costa and Antony Rowstron Microsoft Research, Cambridge, CB3 FB, UK Abstract There has been much interest in both unstructured and structured

More information

Early Measurements of a Cluster-based Architecture for P2P Systems

Early Measurements of a Cluster-based Architecture for P2P Systems Early Measurements of a Cluster-based Architecture for P2P Systems Balachander Krishnamurthy, Jia Wang, Yinglian Xie I. INTRODUCTION Peer-to-peer applications such as Napster [4], Freenet [1], and Gnutella

More information

Towards Location-aware Topology in both Unstructured and Structured P2P Systems

Towards Location-aware Topology in both Unstructured and Structured P2P Systems Towards Location-aware Topology in both Unstructured and Structured P2P Systems Tongqing Qiu, Guihai Chen, Mao Ye State Key Lab of Novel Software Nanjing University Ben Y. Zhao Department of Computer Science

More information

Building a low-latency, proximity-aware DHT-based P2P network

Building a low-latency, proximity-aware DHT-based P2P network Building a low-latency, proximity-aware DHT-based P2P network Ngoc Ben DANG, Son Tung VU, Hoai Son NGUYEN Department of Computer network College of Technology, Vietnam National University, Hanoi 144 Xuan

More information

A Super-Peer Based Lookup in Structured Peer-to-Peer Systems

A Super-Peer Based Lookup in Structured Peer-to-Peer Systems A Super-Peer Based Lookup in Structured Peer-to-Peer Systems Yingwu Zhu Honghao Wang Yiming Hu ECECS Department ECECS Department ECECS Department University of Cincinnati University of Cincinnati University

More information

Application Layer Multicast For Efficient Peer-to-Peer Applications

Application Layer Multicast For Efficient Peer-to-Peer Applications Application Layer Multicast For Efficient Peer-to-Peer Applications Adam Wierzbicki 1 e-mail: adamw@icm.edu.pl Robert Szczepaniak 1 Marcin Buszka 1 1 Polish-Japanese Institute of Information Technology

More information

Simple Determination of Stabilization Bounds for Overlay Networks. are now smaller, faster, and near-omnipresent. Computer ownership has gone from one

Simple Determination of Stabilization Bounds for Overlay Networks. are now smaller, faster, and near-omnipresent. Computer ownership has gone from one Simple Determination of Stabilization Bounds for Overlay Networks A. Introduction The landscape of computing has changed dramatically in the past half-century. Computers are now smaller, faster, and near-omnipresent.

More information

Optimizing Overlay Topology by Reducing Cut Vertices

Optimizing Overlay Topology by Reducing Cut Vertices Optimizing Overlay Topology by Reducing Cut Vertices Xiaomei Liu 1, Li Xiao 1, Andrew Kreling 1, Yunhao Liu 2 1 Department of Computer Science and Engineering, Michigan State University 2 Deptartment of

More information

A Structured Overlay for Non-uniform Node Identifier Distribution Based on Flexible Routing Tables

A Structured Overlay for Non-uniform Node Identifier Distribution Based on Flexible Routing Tables A Structured Overlay for Non-uniform Node Identifier Distribution Based on Flexible Routing Tables Takehiro Miyao, Hiroya Nagao, Kazuyuki Shudo Tokyo Institute of Technology 2-12-1 Ookayama, Meguro-ku,

More information

Peer Clustering and Firework Query Model

Peer Clustering and Firework Query Model Peer Clustering and Firework Query Model Cheuk Hang Ng, Ka Cheung Sia Department of Computer Science and Engineering The Chinese University of Hong Kong Shatin, N.T., Hong Kong SAR {chng,kcsia}@cse.cuhk.edu.hk

More information

AOTO: Adaptive Overlay Topology Optimization in Unstructured P2P Systems

AOTO: Adaptive Overlay Topology Optimization in Unstructured P2P Systems AOTO: Adaptive Overlay Topology Optimization in Unstructured P2P Systems Yunhao Liu, Zhenyun Zhuang, Li Xiao Department of Computer Science and Engineering Michigan State University East Lansing, MI 48824

More information

Implications of Neighbor Selection on DHT Overlays

Implications of Neighbor Selection on DHT Overlays Implications of Neighbor Selection on DHT Overlays Yingwu Zhu Department of CSSE, Seattle University zhuy@seattleu.edu Xiaoyu Yang Department of ECECS, University of Cincinnati yangxu@ececs.uc.edu Abstract

More information

Dynamic Load Sharing in Peer-to-Peer Systems: When some Peers are more Equal than Others

Dynamic Load Sharing in Peer-to-Peer Systems: When some Peers are more Equal than Others Dynamic Load Sharing in Peer-to-Peer Systems: When some Peers are more Equal than Others Sabina Serbu, Silvia Bianchi, Peter Kropf and Pascal Felber Computer Science Department, University of Neuchâtel

More information

A Directed-multicast Routing Approach with Path Replication in Content Addressable Network

A Directed-multicast Routing Approach with Path Replication in Content Addressable Network 2010 Second International Conference on Communication Software and Networks A Directed-multicast Routing Approach with Path Replication in Content Addressable Network Wenbo Shen, Weizhe Zhang, Hongli Zhang,

More information

Effects of Churn on Structured P2P Overlay Networks

Effects of Churn on Structured P2P Overlay Networks International Conference on Automation, Control, Engineering and Computer Science (ACECS'14) Proceedings - Copyright IPCO-214, pp.164-17 ISSN 2356-568 Effects of Churn on Structured P2P Overlay Networks

More information

Characterizing Traffic Demand Aware Overlay Routing Network Topologies

Characterizing Traffic Demand Aware Overlay Routing Network Topologies Characterizing Traffic Demand Aware Overlay Routing Network Topologies Benjamin D. McBride Kansas State University Rathbone Hall Manhattan, KS Email: bdm@ksu.edu Caterina Scoglio Kansas State University

More information

Survey of DHT Evaluation Methods

Survey of DHT Evaluation Methods Survey of DHT Evaluation Methods Markus Meriläinen Helsinki University of Technology Markus.Merilainen@tkk.fi Abstract In this paper, we present an overview of factors affecting the performance of the

More information

A Framework for Peer-To-Peer Lookup Services based on k-ary search

A Framework for Peer-To-Peer Lookup Services based on k-ary search A Framework for Peer-To-Peer Lookup Services based on k-ary search Sameh El-Ansary Swedish Institute of Computer Science Kista, Sweden Luc Onana Alima Department of Microelectronics and Information Technology

More information

Design of a New Hierarchical Structured Peer-to-Peer Network Based On Chinese Remainder Theorem

Design of a New Hierarchical Structured Peer-to-Peer Network Based On Chinese Remainder Theorem Design of a New Hierarchical Structured Peer-to-Peer Network Based On Chinese Remainder Theorem Bidyut Gupta, Nick Rahimi, Henry Hexmoor, and Koushik Maddali Department of Computer Science Southern Illinois

More information

Layer Optimization for DHT-based Peer-to-Peer Network

Layer Optimization for DHT-based Peer-to-Peer Network Layer Optimization for DHT-based Peer-to-Peer Network Jun Li *, Cuilian Li, Zhaoxi Fang Department of Telecommunication Zhejiang Wanli University Ningbo, China xxllj, licl@zwu.edu.cn, zhaoxifang@gmail.com

More information

Time-related replication for p2p storage system

Time-related replication for p2p storage system Seventh International Conference on Networking Time-related replication for p2p storage system Kyungbaek Kim E-mail: University of California, Irvine Computer Science-Systems 3204 Donald Bren Hall, Irvine,

More information

Understanding Chord Performance

Understanding Chord Performance CS68 Course Project Understanding Chord Performance and Topology-aware Overlay Construction for Chord Li Zhuang(zl@cs), Feng Zhou(zf@cs) Abstract We studied performance of the Chord scalable lookup system

More information

EARM: An Efficient and Adaptive File Replication with Consistency Maintenance in P2P Systems.

EARM: An Efficient and Adaptive File Replication with Consistency Maintenance in P2P Systems. : An Efficient and Adaptive File Replication with Consistency Maintenance in P2P Systems. 1 K.V.K.Chaitanya, 2 Smt. S.Vasundra, M,Tech., (Ph.D), 1 M.Tech (Computer Science), 2 Associate Professor, Department

More information

A Peer-to-Peer Architecture to Enable Versatile Lookup System Design

A Peer-to-Peer Architecture to Enable Versatile Lookup System Design A Peer-to-Peer Architecture to Enable Versatile Lookup System Design Vivek Sawant Jasleen Kaur University of North Carolina at Chapel Hill, Chapel Hill, NC, USA vivek, jasleen @cs.unc.edu Abstract The

More information

A Method for Designing Proximity-aware Routing Algorithms for Structured Overlays

A Method for Designing Proximity-aware Routing Algorithms for Structured Overlays A Method for Designing Proximity-aware Routing Algorithms for Structured Overlays Takehiro Miyao, Hiroya Nagao, Kazuyuki Shudo Tokyo Institute of Technology 2-12-1 Ookayama, Meguro-ku, Tokyo, JAPAN Email:

More information

Exploiting Route Redundancy via Structured Peer to Peer Overlays

Exploiting Route Redundancy via Structured Peer to Peer Overlays Exploiting Route Redundancy ia Structured Peer to Peer Oerlays Ben Y. Zhao, Ling Huang, Jeremy Stribling, Anthony D. Joseph, and John D. Kubiatowicz Uniersity of California, Berkeley Challenges Facing

More information

Scalability In Peer-to-Peer Systems. Presented by Stavros Nikolaou

Scalability In Peer-to-Peer Systems. Presented by Stavros Nikolaou Scalability In Peer-to-Peer Systems Presented by Stavros Nikolaou Background on Peer-to-Peer Systems Definition: Distributed systems/applications featuring: No centralized control, no hierarchical organization

More information

Distributed Hash Table

Distributed Hash Table Distributed Hash Table P2P Routing and Searching Algorithms Ruixuan Li College of Computer Science, HUST rxli@public.wh.hb.cn http://idc.hust.edu.cn/~rxli/ In Courtesy of Xiaodong Zhang, Ohio State Univ

More information

DYNAMIC TREE-LIKE STRUCTURES IN P2P-NETWORKS

DYNAMIC TREE-LIKE STRUCTURES IN P2P-NETWORKS DYNAMIC TREE-LIKE STRUCTURES IN P2P-NETWORKS Herwig Unger Markus Wulff Department of Computer Science University of Rostock D-1851 Rostock, Germany {hunger,mwulff}@informatik.uni-rostock.de KEYWORDS P2P,

More information

A Routing Mechanism by Distance-weighted Bloom Filter *

A Routing Mechanism by Distance-weighted Bloom Filter * 7659, England, UK Journal of Information and Computing Science Vol. 2, No. 1, 2007, pp. 55-60 A Routing Mechanism by Distance-weighted Bloom Filter * Xun Duan + and Jian-shi Li School of Computer Science

More information

Peer-to-Peer Systems. Chapter General Characteristics

Peer-to-Peer Systems. Chapter General Characteristics Chapter 2 Peer-to-Peer Systems Abstract In this chapter, a basic overview is given of P2P systems, architectures, and search strategies in P2P systems. More specific concepts that are outlined include

More information

Virtual Multi-homing: On the Feasibility of Combining Overlay Routing with BGP Routing

Virtual Multi-homing: On the Feasibility of Combining Overlay Routing with BGP Routing Virtual Multi-homing: On the Feasibility of Combining Overlay Routing with BGP Routing Zhi Li, Prasant Mohapatra, and Chen-Nee Chuah University of California, Davis, CA 95616, USA {lizhi, prasant}@cs.ucdavis.edu,

More information

UC Berkeley UC Berkeley Previously Published Works

UC Berkeley UC Berkeley Previously Published Works UC Berkeley UC Berkeley Previously Published Works Title Impact of neighbor selection on performance and resilience of structured P2P networks Permalink https://escholarship.org/uc/item/9tq2wn2 Authors

More information

An Adaptive Stabilization Framework for DHT

An Adaptive Stabilization Framework for DHT An Adaptive Stabilization Framework for DHT Gabriel Ghinita, Yong Meng Teo Department of Computer Science National University of Singapore {ghinitag,teoym}@comp.nus.edu.sg Overview Background Related Work

More information

A Hybrid Peer-to-Peer Architecture for Global Geospatial Web Service Discovery

A Hybrid Peer-to-Peer Architecture for Global Geospatial Web Service Discovery A Hybrid Peer-to-Peer Architecture for Global Geospatial Web Service Discovery Shawn Chen 1, Steve Liang 2 1 Geomatics, University of Calgary, hschen@ucalgary.ca 2 Geomatics, University of Calgary, steve.liang@ucalgary.ca

More information

A SDN-like Loss Recovery Solution in Application Layer Multicast Wenqing Lei 1, Cheng Ma 1, Xinchang Zhang 2, a, Lu Wang 2

A SDN-like Loss Recovery Solution in Application Layer Multicast Wenqing Lei 1, Cheng Ma 1, Xinchang Zhang 2, a, Lu Wang 2 5th International Conference on Information Engineering for Mechanics and Materials (ICIMM 2015) A SDN-like Loss Recovery Solution in Application Layer Multicast Wenqing Lei 1, Cheng Ma 1, Xinchang Zhang

More information

PChord: Improvement on Chord to Achieve Better Routing Efficiency by Exploiting Proximity

PChord: Improvement on Chord to Achieve Better Routing Efficiency by Exploiting Proximity 546 PAPER Special Section on Parallel/Distributed Computing and Networking PChord: Improvement on Chord to Achieve Better Routing Efficiency by Exploiting Proximity Feng HONG a),mingluli,minyouwu, and

More information

Thwarting Traceback Attack on Freenet

Thwarting Traceback Attack on Freenet Thwarting Traceback Attack on Freenet Guanyu Tian, Zhenhai Duan Florida State University {tian, duan}@cs.fsu.edu Todd Baumeister, Yingfei Dong University of Hawaii {baumeist, yingfei}@hawaii.edu Abstract

More information

LessLog: A Logless File Replication Algorithm for Peer-to-Peer Distributed Systems

LessLog: A Logless File Replication Algorithm for Peer-to-Peer Distributed Systems LessLog: A Logless File Replication Algorithm for Peer-to-Peer Distributed Systems Kuang-Li Huang, Tai-Yi Huang and Jerry C. Y. Chou Department of Computer Science National Tsing Hua University Hsinchu,

More information

Distriubted Hash Tables and Scalable Content Adressable Network (CAN)

Distriubted Hash Tables and Scalable Content Adressable Network (CAN) Distriubted Hash Tables and Scalable Content Adressable Network (CAN) Ines Abdelghani 22.09.2008 Contents 1 Introduction 2 2 Distributed Hash Tables: DHT 2 2.1 Generalities about DHTs............................

More information

Defending against Eclipse attacks on overlay networks

Defending against Eclipse attacks on overlay networks Defending against Eclipse attacks on overlay networks Atul Singh 1 Miguel Castro 2 Peter Druschel 1 Antony Rowstron 2 1 Rice University, Houston, TX, USA. 2 Microsoft Research, Cambridge, UK. Abstract

More information

MULTI-DOMAIN VoIP PEERING USING OVERLAY NETWORK

MULTI-DOMAIN VoIP PEERING USING OVERLAY NETWORK 116 MULTI-DOMAIN VoIP PEERING USING OVERLAY NETWORK Herry Imanta Sitepu, Carmadi Machbub, Armein Z. R. Langi, Suhono Harso Supangkat School of Electrical Engineering and Informatics, Institut Teknologi

More information

A Top Catching Scheme Consistency Controlling in Hybrid P2P Network

A Top Catching Scheme Consistency Controlling in Hybrid P2P Network A Top Catching Scheme Consistency Controlling in Hybrid P2P Network V. Asha*1, P Ramesh Babu*2 M.Tech (CSE) Student Department of CSE, Priyadarshini Institute of Technology & Science, Chintalapudi, Guntur(Dist),

More information

Effect of Links on DHT Routing Algorithms 1

Effect of Links on DHT Routing Algorithms 1 Effect of Links on DHT Routing Algorithms 1 Futai Zou, Liang Zhang, Yin Li, Fanyuan Ma Department of Computer Science and Engineering Shanghai Jiao Tong University, 200030 Shanghai, China zoufutai@cs.sjtu.edu.cn

More information

A Search Theoretical Approach to P2P Networks: Analysis of Learning

A Search Theoretical Approach to P2P Networks: Analysis of Learning A Search Theoretical Approach to P2P Networks: Analysis of Learning Nazif Cihan Taş Dept. of Computer Science University of Maryland College Park, MD 2742 Email: ctas@cs.umd.edu Bedri Kâmil Onur Taş Dept.

More information

Performance Modelling of Peer-to-Peer Routing

Performance Modelling of Peer-to-Peer Routing Performance Modelling of Peer-to-Peer Routing Idris A. Rai, Andrew Brampton, Andrew MacQuire and Laurent Mathy Computing Department, Lancaster University {rai,brampton,macquire,laurent}@comp.lancs.ac.uk

More information

Comparing Chord, CAN, and Pastry Overlay Networks for Resistance to DoS Attacks

Comparing Chord, CAN, and Pastry Overlay Networks for Resistance to DoS Attacks Comparing Chord, CAN, and Pastry Overlay Networks for Resistance to DoS Attacks Hakem Beitollahi Hakem.Beitollahi@esat.kuleuven.be Geert Deconinck Geert.Deconinck@esat.kuleuven.be Katholieke Universiteit

More information

SplitQuest: Controlled and Exhaustive Search in Peer-to-Peer Networks

SplitQuest: Controlled and Exhaustive Search in Peer-to-Peer Networks SplitQuest: Controlled and Exhaustive Search in Peer-to-Peer Networks Pericles Lopes Ronaldo A. Ferreira pericles@facom.ufms.br raf@facom.ufms.br College of Computing, Federal University of Mato Grosso

More information

Shaking Service Requests in Peer-to-Peer Video Systems

Shaking Service Requests in Peer-to-Peer Video Systems Service in Peer-to-Peer Video Systems Ying Cai Ashwin Natarajan Johnny Wong Department of Computer Science Iowa State University Ames, IA 500, U. S. A. E-mail: {yingcai, ashwin, wong@cs.iastate.edu Abstract

More information

Peer-to-peer systems and overlay networks

Peer-to-peer systems and overlay networks Complex Adaptive Systems C.d.L. Informatica Università di Bologna Peer-to-peer systems and overlay networks Fabio Picconi Dipartimento di Scienze dell Informazione 1 Outline Introduction to P2P systems

More information

Challenges in the Wide-area. Tapestry: Decentralized Routing and Location. Global Computation Model. Cluster-based Applications

Challenges in the Wide-area. Tapestry: Decentralized Routing and Location. Global Computation Model. Cluster-based Applications Challenges in the Wide-area Tapestry: Decentralized Routing and Location System Seminar S 0 Ben Y. Zhao CS Division, U. C. Berkeley Trends: Exponential growth in CPU, b/w, storage Network expanding in

More information

Hybrid Overlay Structure Based on Random Walks

Hybrid Overlay Structure Based on Random Walks Hybrid Overlay Structure Based on Random Walks Ruixiong Tian 1,, Yongqiang Xiong 2, Qian Zhang 2,BoLi 3, Ben Y. Zhao 4, and Xing Li 1 1 Department of Electronic Engineering, Tsinghua University 2 Microsoft

More information

Evaluating Unstructured Peer-to-Peer Lookup Overlays

Evaluating Unstructured Peer-to-Peer Lookup Overlays Evaluating Unstructured Peer-to-Peer Lookup Overlays Idit Keidar EE Department, Technion Roie Melamed CS Department, Technion ABSTRACT Unstructured peer-to-peer lookup systems incur small constant overhead

More information

Lecture 6: Overlay Networks. CS 598: Advanced Internetworking Matthew Caesar February 15, 2011

Lecture 6: Overlay Networks. CS 598: Advanced Internetworking Matthew Caesar February 15, 2011 Lecture 6: Overlay Networks CS 598: Advanced Internetworking Matthew Caesar February 15, 2011 1 Overlay networks: Motivations Protocol changes in the network happen very slowly Why? Internet is shared

More information

A DHT-Based Grid Resource Indexing and Discovery Scheme

A DHT-Based Grid Resource Indexing and Discovery Scheme SINGAPORE-MIT ALLIANCE SYMPOSIUM 2005 1 A DHT-Based Grid Resource Indexing and Discovery Scheme Yong Meng TEO 1,2, Verdi March 2 and Xianbing Wang 1 1 Singapore-MIT Alliance, 2 Department of Computer Science,

More information

Decentralized Object Location In Dynamic Peer-to-Peer Distributed Systems

Decentralized Object Location In Dynamic Peer-to-Peer Distributed Systems Decentralized Object Location In Dynamic Peer-to-Peer Distributed Systems George Fletcher Project 3, B649, Dr. Plale July 16, 2003 1 Introduction One of the key requirements for global level scalability

More information

Searching for Shared Resources: DHT in General

Searching for Shared Resources: DHT in General 1 ELT-53206 Peer-to-Peer Networks Searching for Shared Resources: DHT in General Mathieu Devos Tampere University of Technology Department of Electronics and Communications Engineering Based on the original

More information

Plover: A Proactive Low-overhead File Replication Scheme for Structured P2P Systems

Plover: A Proactive Low-overhead File Replication Scheme for Structured P2P Systems : A Proactive Low-overhead File Replication Scheme for Structured P2P Systems Haiying Shen Yingwu Zhu Dept. of Computer Science & Computer Engineering Dept. of Computer Science & Software Engineering University

More information

Overlay Networks for Multimedia Contents Distribution

Overlay Networks for Multimedia Contents Distribution Overlay Networks for Multimedia Contents Distribution Vittorio Palmisano vpalmisano@gmail.com 26 gennaio 2007 Outline 1 Mesh-based Multicast Networks 2 Tree-based Multicast Networks Overcast (Cisco, 2000)

More information

March 10, Distributed Hash-based Lookup. for Peer-to-Peer Systems. Sandeep Shelke Shrirang Shirodkar MTech I CSE

March 10, Distributed Hash-based Lookup. for Peer-to-Peer Systems. Sandeep Shelke Shrirang Shirodkar MTech I CSE for for March 10, 2006 Agenda for Peer-to-Peer Sytems Initial approaches to Their Limitations CAN - Applications of CAN Design Details Benefits for Distributed and a decentralized architecture No centralized

More information

Relaxing Routing Table to Alleviate Dynamism in P2P Systems

Relaxing Routing Table to Alleviate Dynamism in P2P Systems Relaxing Routing Table to Alleviate Dynamism in P2P Systems Hui FANG 1, Wen Jing HSU 2, and Larry RUDOLPH 3 1 Singapore-MIT Alliance, National University of Singapore 2 Nanyang Technological University,

More information

Routing Table Construction Method Solely Based on Query Flows for Structured Overlays

Routing Table Construction Method Solely Based on Query Flows for Structured Overlays Routing Table Construction Method Solely Based on Query Flows for Structured Overlays Yasuhiro Ando, Hiroya Nagao, Takehiro Miyao and Kazuyuki Shudo Tokyo Institute of Technology Abstract In structured

More information

Searching for Shared Resources: DHT in General

Searching for Shared Resources: DHT in General 1 ELT-53207 P2P & IoT Systems Searching for Shared Resources: DHT in General Mathieu Devos Tampere University of Technology Department of Electronics and Communications Engineering Based on the original

More information

ReCord: A Distributed Hash Table with Recursive Structure

ReCord: A Distributed Hash Table with Recursive Structure ReCord: A Distributed Hash Table with Recursive Structure Jianyang Zeng and Wen-Jing Hsu Abstract We propose a simple distributed hash table called ReCord, which is a generalized version of Randomized-

More information

A Chord-Based Novel Mobile Peer-to-Peer File Sharing Protocol

A Chord-Based Novel Mobile Peer-to-Peer File Sharing Protocol A Chord-Based Novel Mobile Peer-to-Peer File Sharing Protocol Min Li 1, Enhong Chen 1, and Phillip C-y Sheu 2 1 Department of Computer Science and Technology, University of Science and Technology of China,

More information

DISTRIBUTED HASH TABLE PROTOCOL DETECTION IN WIRELESS SENSOR NETWORKS

DISTRIBUTED HASH TABLE PROTOCOL DETECTION IN WIRELESS SENSOR NETWORKS DISTRIBUTED HASH TABLE PROTOCOL DETECTION IN WIRELESS SENSOR NETWORKS Mr. M. Raghu (Asst.professor) Dr.Pauls Engineering College Ms. M. Ananthi (PG Scholar) Dr. Pauls Engineering College Abstract- Wireless

More information

Efficient Multi-source Data Dissemination in Peer-to-Peer Networks

Efficient Multi-source Data Dissemination in Peer-to-Peer Networks Efficient Multi-source Data Dissemination in Peer-to-Peer Networks Zhenyu Li 1,2, Zengyang Zhu 1,2, Gaogang Xie 1, Zhongcheng Li 1 1 Institute of Computing Technology, Chinese Academy of Sciences 2 Graduate

More information

Dorina Luminiţa COPACI, Constantin Alin COPACI

Dorina Luminiţa COPACI, Constantin Alin COPACI THE DESIGN OF RESILIENCE P2P NETWORKS WITH DISTRIBUTED HASH TABLES Dorina Luminiţa COPACI, Constantin Alin COPACI lcopaci@yahoo.com, acopaci@yahoo.com Abstract The term resilience in computer systems and

More information

Update Propagation Through Replica Chain in Decentralized and Unstructured P2P Systems

Update Propagation Through Replica Chain in Decentralized and Unstructured P2P Systems Update Propagation Through Replica Chain in Decentralized and Unstructured PP Systems Zhijun Wang, Sajal K. Das, Mohan Kumar and Huaping Shen Center for Research in Wireless Mobility and Networking (CReWMaN)

More information

Problems in Reputation based Methods in P2P Networks

Problems in Reputation based Methods in P2P Networks WDS'08 Proceedings of Contributed Papers, Part I, 235 239, 2008. ISBN 978-80-7378-065-4 MATFYZPRESS Problems in Reputation based Methods in P2P Networks M. Novotný Charles University, Faculty of Mathematics

More information

Debunking some myths about structured and unstructured overlays

Debunking some myths about structured and unstructured overlays Debunking some myths about structured and unstructured overlays Miguel Castro Manuel Costa Antony Rowstron Microsoft Research, 7 J J Thomson Avenue, Cambridge, UK Abstract We present a comparison of structured

More information

Handling Churn in a DHT

Handling Churn in a DHT Handling Churn in a DHT Sean Rhea, Dennis Geels, Timothy Roscoe, and John Kubiatowicz UC Berkeley and Intel Research Berkeley What s a DHT? Distributed Hash Table Peer-to-peer algorithm to offering put/get

More information

Comparing the performance of distributed hash tables under churn

Comparing the performance of distributed hash tables under churn Comparing the performance of distributed hash tables under churn Jinyang Li, Jeremy Stribling, Thomer M. Gil, Robert Morris, M. Frans Kaashoek MIT Computer Science and Artificial Intelligence Laboratory

More information

BOOTSTRAPPING LOCALITY-AWARE P2P NETWORKS

BOOTSTRAPPING LOCALITY-AWARE P2P NETWORKS BOOTSTRAPPING LOCALITY-AWARE PP NETWORKS Curt Cramer, Kendy Kutzner, and Thomas Fuhrmann Institut für Telematik, Universität Karlsruhe (TH), Germany {curt.cramer kendy.kutzner thomas.fuhrmann}@ira.uka.de

More information

Towards Scalable and Robust Overlay Networks

Towards Scalable and Robust Overlay Networks Towards Scalable and Robust Overlay Networks Baruch Awerbuch Department of Computer Science Johns Hopkins University Baltimore, MD 21218, USA baruch@cs.jhu.edu Christian Scheideler Institute for Computer

More information

08 Distributed Hash Tables

08 Distributed Hash Tables 08 Distributed Hash Tables 2/59 Chord Lookup Algorithm Properties Interface: lookup(key) IP address Efficient: O(log N) messages per lookup N is the total number of servers Scalable: O(log N) state per

More information

Impact of Neighbor Selection on Performance and Resilience of Structured P2P Networks

Impact of Neighbor Selection on Performance and Resilience of Structured P2P Networks Impact of Neighbor Selection on Performance and Resilience of Structured P2P Networks Byung-Gon Chun, Ben Y. Zhao 2, and John D. Kubiatowicz Computer Science Division, U.C. Berkeley {bgchun, kubitron}@cs.berkeley.edu

More information

Brocade: Landmark Routing on Overlay Networks

Brocade: Landmark Routing on Overlay Networks Abstract Brocade: Landmark Routing on Overlay Networks CS262A Fall 2001 Yitao Duan, Ling Huang University of California, Berkeley duan@cs.berkeley.edu, hlion@newton.berkeley.edu Peer-to-peer networks offer

More information

A Scalable Content- Addressable Network

A Scalable Content- Addressable Network A Scalable Content- Addressable Network In Proceedings of ACM SIGCOMM 2001 S. Ratnasamy, P. Francis, M. Handley, R. Karp, S. Shenker Presented by L.G. Alex Sung 9th March 2005 for CS856 1 Outline CAN basics

More information

Characterizing Gnutella Network Properties for Peer-to-Peer Network Simulation

Characterizing Gnutella Network Properties for Peer-to-Peer Network Simulation Characterizing Gnutella Network Properties for Peer-to-Peer Network Simulation Selim Ciraci, Ibrahim Korpeoglu, and Özgür Ulusoy Department of Computer Engineering, Bilkent University, TR-06800 Ankara,

More information

Bridge-Node Selection and Loss Recovery in Island Multicast

Bridge-Node Selection and Loss Recovery in Island Multicast Bridge-Node Selection and Loss Recovery in Island Multicast W.-P. Ken Yiu K.-F. Simon Wong S.-H. Gary Chan Department of Computer Science The Hong Kong University of Science and Technology Clear Water

More information

Graph Algorithms. Many problems in networks can be modeled as graph problems.

Graph Algorithms. Many problems in networks can be modeled as graph problems. Graph Algorithms Graph Algorithms Many problems in networks can be modeled as graph problems. - The topology of a distributed system is a graph. - Routing table computation uses the shortest path algorithm

More information

Back-Up Chord: Chord Ring Recovery Protocol for P2P File Sharing over MANETs

Back-Up Chord: Chord Ring Recovery Protocol for P2P File Sharing over MANETs Back-Up Chord: Chord Ring Recovery Protocol for P2P File Sharing over MANETs Hong-Jong Jeong, Dongkyun Kim, Jeomki Song, Byung-yeub Kim, and Jeong-Su Park Department of Computer Engineering, Kyungpook

More information

Diminished Chord: A Protocol for Heterogeneous Subgroup Formation in Peer-to-Peer Networks

Diminished Chord: A Protocol for Heterogeneous Subgroup Formation in Peer-to-Peer Networks Diminished Chord: A Protocol for Heterogeneous Subgroup Formation in Peer-to-Peer Networks David R. Karger 1 and Matthias Ruhl 2 1 MIT Computer Science and Artificial Intelligence Laboratory Cambridge,

More information

Architectures for Distributed Systems

Architectures for Distributed Systems Distributed Systems and Middleware 2013 2: Architectures Architectures for Distributed Systems Components A distributed system consists of components Each component has well-defined interface, can be replaced

More information

Protocol for Tetherless Computing

Protocol for Tetherless Computing Protocol for Tetherless Computing S. Keshav P. Darragh A. Seth S. Fung School of Computer Science University of Waterloo Waterloo, Canada, N2L 3G1 1. Introduction Tetherless computing involves asynchronous

More information

Challenges in the Wide-area. Tapestry: Decentralized Routing and Location. Key: Location and Routing. Driving Applications

Challenges in the Wide-area. Tapestry: Decentralized Routing and Location. Key: Location and Routing. Driving Applications Challenges in the Wide-area Tapestry: Decentralized Routing and Location SPAM Summer 00 Ben Y. Zhao CS Division, U. C. Berkeley! Trends: Exponential growth in CPU, b/w, storage Network expanding in reach

More information

Introduction to Peer-to-Peer Systems

Introduction to Peer-to-Peer Systems Introduction Introduction to Peer-to-Peer Systems Peer-to-peer (PP) systems have become extremely popular and contribute to vast amounts of Internet traffic PP basic definition: A PP system is a distributed

More information

Peer-To-Peer Techniques

Peer-To-Peer Techniques PG DynaSearch Markus Benter 31th October, 2013 Introduction Centralized P2P-Networks Unstructured P2P-Networks Structured P2P-Networks Misc 1 What is a Peer-to-Peer System? Definition Peer-to-peer systems

More information

Minimizing Churn in Distributed Systems

Minimizing Churn in Distributed Systems Minimizing Churn in Distributed Systems by P. Brighten Godfrey, Scott Shenker, and Ion Stoica appearing in SIGCOMM 2006 presented by Todd Sproull Introduction Problem: nodes joining or leaving distributed

More information

Overlay and P2P Networks. Introduction and unstructured networks. Prof. Sasu Tarkoma

Overlay and P2P Networks. Introduction and unstructured networks. Prof. Sasu Tarkoma Overlay and P2P Networks Introduction and unstructured networks Prof. Sasu Tarkoma 14.1.2013 Contents Overlay networks and intro to networking Unstructured networks Overlay Networks An overlay network

More information

Load Sharing in Peer-to-Peer Networks using Dynamic Replication

Load Sharing in Peer-to-Peer Networks using Dynamic Replication Load Sharing in Peer-to-Peer Networks using Dynamic Replication S Rajasekhar, B Rong, K Y Lai, I Khalil and Z Tari School of Computer Science and Information Technology RMIT University, Melbourne 3, Australia

More information

Scalable P2P architectures

Scalable P2P architectures Scalable P2P architectures Oscar Boykin Electrical Engineering, UCLA Joint work with: Jesse Bridgewater, Joseph Kong, Kamen Lozev, Behnam Rezaei, Vwani Roychowdhury, Nima Sarshar Outline Introduction to

More information

Exploiting the Synergy between Peer-to-Peer and Mobile Ad Hoc Networks

Exploiting the Synergy between Peer-to-Peer and Mobile Ad Hoc Networks Exploiting the Synergy between Peer-to-Peer and Mobile Ad Hoc Networks Y. Charlie Hu, Saumitra M. Das, and Himabindu Pucha Purdue University West Lafayette, IN 47907 {ychu, smdas, hpucha}@purdue.edu Abstract

More information

Supplementary file for SybilDefender: A Defense Mechanism for Sybil Attacks in Large Social Networks

Supplementary file for SybilDefender: A Defense Mechanism for Sybil Attacks in Large Social Networks 1 Supplementary file for SybilDefender: A Defense Mechanism for Sybil Attacks in Large Social Networks Wei Wei, Fengyuan Xu, Chiu C. Tan, Qun Li The College of William and Mary, Temple University {wwei,

More information

Cycloid: A constant-degree and lookup-efficient P2P overlay network

Cycloid: A constant-degree and lookup-efficient P2P overlay network Performance Evaluation xxx (2005) xxx xxx Cycloid: A constant-degree and lookup-efficient P2P overlay network Haiying Shen a, Cheng-Zhong Xu a,, Guihai Chen b a Department of Electrical and Computer Engineering,

More information

Peer-to-Peer Systems. Network Science: Introduction. P2P History: P2P History: 1999 today

Peer-to-Peer Systems. Network Science: Introduction. P2P History: P2P History: 1999 today Network Science: Peer-to-Peer Systems Ozalp Babaoglu Dipartimento di Informatica Scienza e Ingegneria Università di Bologna www.cs.unibo.it/babaoglu/ Introduction Peer-to-peer (PP) systems have become

More information

Performance Analysis of Peer-to-Peer Networks for File Distribution

Performance Analysis of Peer-to-Peer Networks for File Distribution Performance Analysis of Peer-to-Peer Networks for File Distribution Ernst W. Biersack, Pablo Rodriguez, and Pascal Felber Institut EURECOM, France {erbi,felber}@eurecom.fr Microsoft Research, UK pablo@microsoft.com

More information