Techniques for improving predictability and message efficiency of gossip protocols

Techniques for Improving Predictability and Message Efficiency of Gossip Protocols SATISH KUMAR VERMA B.Tech., IIT Madras A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY SCHOOL OF COMPUTING NATIONAL UNIVERSITY OF SINGAPORE 2008-2009 Acknowledgements First of all, I would like to thank my advisor Dr. Ooi Wei Tsang, without whose guidance, both academic and as a friend, I could not have completed my Ph.D. research. Particularly, I owe him for helping me recognize the significance of understanding any phenomenon through analytical modeling, and presenting results in a precise manner. Though, I still have a lot to learn. I also wish to thank my thesis committee members, Dr. Chan Mun Choon and Dr. Gary Tan for their patience and comments throughout the duration of my research. I would also like to express my gratitude to the faculty of School of Computing for sharing their knowledge. In addition, I wish to thank the staff of School of Computing for helping with any matter I needed help with. I also wish to thank NUS for the generous scholarship and excellent infrastructure for work and life. I cherish the time together with my fellow lab mates: Gu Yan, Cheng Wei, Ma Lin, Raman Balaji, Dan Liu, Pavel Korshunov, Hemendra Singh Negi and Navendu Singh. Their constant encouragement and friendship made the long journey enjoyable. Last but not the least, I would like to thank Maricar, for sharing my joy and sadness, and for giving her sweet and patient love during this long ordeal. Finally, I am forever indebted to my family for supporting me always. i Table of Contents Introduction 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Gossip: Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Approaches to Large-scale Information Dissemination . . . . . . . . . . . . . . . . 1.3.1 Unicast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.2 Deterministic Tree/Mesh-based Multicast . . . . . . . . . . . . . . . . . . . 1.3.3 Randomized Gossip Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Comparing Deterministic Approaches and Randomized Gossip . . . . . . . . . . . 1.4.1 Scalability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2 Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.3 Fault-tolerance and Robustness . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.4 Trade-offs in Using Gossip . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Gossip: Key Problems Addressed in the Thesis . . . . . . . . . . . . . . . . . . . . 1.5.1 Randomness in Latency of Delivery . . . . . . . . . . . . . . . . . . . . . . 1.5.2 High Transmission Overhead . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 List of Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.1 Fine-grained Control of Gossip Protocol Infection Pattern Using Adaptive Fanout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.2 Hierarchical Extension to Asynchronous Gossip for Better and more Predictable Latency Performance . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.3 Rateless Gossip: Push Gossip with Rateless Codes to reduce Transmission Overhead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7 Structure of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background and Related Work 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Gossip Protocols: Models . . . . . . . . . . . . . . . . . 2.2.1 Process States during a Gossip Protocol . . . . . 2.2.2 Anti-Entropy . . . . . . . . . . . . . . . . . . . . 2.2.3 Rumor-Mongering . . . . . . . . . . . . . . . . . 2.2.4 Aggregate Computing Gossip Protocols . . . . . 2.2.5 Random Phone Call Model . . . . . . . . . . . . 2.2.6 Topology Aware Gossip and Hierarchical Gossip 2.2.7 Push and Pull Gossip . . . . . . . . . . . . . . . 2.2.8 Uniform and Spatial Gossip . . . . . . . . . . . . ii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 6 8 9 10 10 11 12 13 14 14 15 16 17 17 18 19 20 2.3 2.4 2.5 2.6 2.7 2.2.9 Address-dependent and Address-independent Gossip Protocols . . . . . 2.2.10 Implementation of Gossip . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.11 Theoretical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gossip Protocols: Design Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1 Round Based Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2 Fanout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.3 Topology Awareness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.4 Application Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.5 Membership Information . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.6 Push or Pull . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.7 Implication of Message Size . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.8 Issues of Robustness Against Failures . . . . . . . . . . . . . . . . . . . 2.3.9 Other Design Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gossip Protocols: Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 Gossip as a Design Paradigm to Counter Stochastic Scalability Limits . 2.4.2 Large-Scale Information Dissemination . . . . . . . . . . . . . . . . . . . 2.4.3 Gossip-based Failure Detector . . . . . . . . . . . . . . . . . . . . . . . . 2.4.4 Gossip Style Garbage Collection Scheme . . . . . . . . . . . . . . . . . . 2.4.5 Gossip for Resource Location Problem . . . . . . . . . . . . . . . . . . . 2.4.6 Gossip-based Group Membership . . . . . . . . . . . . . . . . . . . . . . 2.4.7 Gossip-Based Algorithms for DB Replicas State Consistency . . . . . . 2.4.8 Gossip Applications in Wireless and Sensor Networks . . . . . . . . . . 2.4.9 Gossip Applications in P2P Networks . . . . . . . . . . . . . . . . . . . 2.4.10 Other Gossip Applications . . . . . . . . . . . . . . . . . . . . . . . . . . Relationship to Our Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.1 Push Gossip Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.2 Flat Gossip Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.3 Synchronous Gossip Model . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.4 Membership Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.5 Fanout as Design Parameter . . . . . . . . . . . . . . . . . . . . . . . . 2.5.6 Application Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.7 Related Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Map of our Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6.1 Addressing High and Random Latency of Data Delivery in Push Gossip 2.6.2 Extension to Hierarchical Gossip . . . . . . . . . . . . . . . . . . . . . . 2.6.3 Address High Message Overhead . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Controlling Gossip Protocol Infection Pattern Using Adaptive Fanout 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Network Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Research Objective and Preliminaries . . . . . . . . . . . . . . . . . . . . . 3.4 Synchronous Gossip Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Fanout for Synchronous Gossip Model . . . . . . . . . . . . . . . . . 3.4.2 Hop-based Interpretation of Synchronous Gossip Model . . . . . . . iii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 21 22 23 24 24 25 26 26 27 28 28 28 29 29 30 31 31 32 33 33 34 35 35 36 37 37 38 38 38 39 39 39 39 40 40 40 . . . . . . 41 41 42 43 46 47 48 3.5 3.6 3.7 3.8 3.9 Interpreting Time in PseudoSynchronous and Asynchronous Gossip Models . . . . 3.5.1 Hop Progress as a Function of Time . . . . . . . . . . . . . . . . . . . . . . PseudoSynchronous Gossip Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.1 Time-based HopContribution Equation for PseudoSynchronous Protocol . 3.6.2 Obtaining Hop-based Fanout for PseudoSynchronous Gossip from User Input Pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Asynchronous Gossip Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7.1 Time-based Fanout for Asynchronous Protocol . . . . . . . . . . . . . . . . 3.7.2 HopContribution Values in Asynchronous Protocol . . . . . . . . . . . . . . Simulations Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.1 Results on Synchronous Gossip . . . . . . . . . . . . . . . . . . . . . . . . . 3.8.2 Results on Asynchronous Gossip . . . . . . . . . . . . . . . . . . . . . . . . Summary and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hierarchical Gossip 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Network Coordinates . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Clustering Approaches in Internet . . . . . . . . . . . . . . 4.2.3 Network Coordinates on Internet . . . . . . . . . . . . . . . 4.3 Hierarchical Gossip Protocol . . . . . . . . . . . . . . . . . . . . . . 4.3.1 K-means Clustering . . . . . . . . . . . . . . . . . . . . . . 4.3.2 Clustering Protocol . . . . . . . . . . . . . . . . . . . . . . . 4.3.3 Cluster Leaders . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.4 Membership Information . . . . . . . . . . . . . . . . . . . . 4.3.5 Gossip Protocol . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.6 Advantages of Hierarchical Gossip . . . . . . . . . . . . . . 4.4 Implementing Asynchronous Gossip in Hierarchical Gossip . . . . . 4.5 Experiments on PlanetLab . . . . . . . . . . . . . . . . . . . . . . . 4.5.1 Clustering of PlanetLab Nodes . . . . . . . . . . . . . . . . 4.5.2 Computing Asynchronous Parameters for Global Gossip . . 4.5.3 Computing Asynchronous Parameters for Various Clusters . 4.5.4 Latency and Message Performance of Hierarchical Gossip vs 4.5.5 Predictability of Hierarchical Gossip vs Global Gossip . . . 4.6 Summary and Future Work . . . . . . . . . . . . . . . . . . . . . . Rateless Gossip: Push Gossip with Rateless Codes 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 5.2 Related Work . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Application Layer Multicast . . . . . . . . . . 5.2.2 Gossip and Message Overhead . . . . . . . . 5.2.3 Network Coding . . . . . . . . . . . . . . . . 5.2.4 Rateless Codes . . . . . . . . . . . . . . . . . 5.3 Analysis of Push Gossip . . . . . . . . . . . . . . . . 5.3.1 Im as a function of m . . . . . . . . . . . . . iv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Global Gossip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 51 51 54 55 56 58 60 62 62 63 66 68 69 70 71 72 74 75 76 77 78 78 78 80 81 82 84 85 87 89 91 94 98 99 102 102 103 103 104 105 107 5.3.2 Computation of P r[0 m i] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 h 5.4 5.5 5.6 5.7 5.3.3 Computating P r[0 i] . . . . . . . . . . . . Rateless Gossip . . . . . . . . . . . . . . . . . . . . 5.4.1 Analysis of Rateless Gossip . . . . . . . . . 5.4.2 Decoding Probability . . . . . . . . . . . . . 5.4.3 Message Distributions . . . . . . . . . . . . 5.4.4 Computing Lθ . . . . . . . . . . . . . . . . Optimized Rateless Gossip . . . . . . . . . . . . . . 5.5.1 α for Optimized Rateless Gossip . . . . . . 5.5.2 Analysis of Optimized Rateless Gossip . . . Simulation Results and Discussion . . . . . . . . . 5.6.1 Push Gossip . . . . . . . . . . . . . . . . . . 5.6.2 Rateless Gossip . . . . . . . . . . . . . . . . 5.6.3 Performance of Optimized Rateless Gossip . 5.6.4 Source Overhead . . . . . . . . . . . . . . . Summary and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 117 119 119 120 122 122 124 125 127 128 128 130 132 135 Conclusions and Future Work 137 6.1 Gossip Protocols with Predictable Behavior over Time . . . . . . . . . . . . . . . . 137 6.2 Hierarchical Gossip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 6.3 Rateless Gossip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 v Abstract Techniques for Improving Predictability and Message Efficiency of Gossip Protocols Satish Kumar Verma National University of Singapore Gossip-based protocols are a class of randomized probabilistic algorithms which offer an attractive design paradigm for large-scale distributed systems. Gossip protocols draw their basic inspiration from a special branch of mathematics, epidemiology, which studies the spread of epidemics in the real world, and hence, are also referred to as Epidemic protocols. Gossip protocols lend themselves to the probabilistic modeling of epidemiological processes. Gossip protocols have gained prominence as an interesting and pragmatic protocol design approach for large systems where the critical challenges, which the conventional deterministic protocols fail to address effectively, are those of scalability, reliability, fault-tolerance, stable throughput, and robustness to system dynamics. A gossip-based communication protocol simply means that: in each step, nodes exchange messages with other nodes which are randomly picked from the respective nodes’ membership view, and over a sequence of such steps, the messages spread throughout the system with high probability, just like an epidemic spreads, from one to another and so on. In this dissertation, we tackle two fundamental challenges faced by gossip algorithms, and propose techniques to improve the efficiency and performance of gossip protocols. The first challenge we tackle is the high and random latency of data delivery that gossip protocols incur. The conventional model to analyze gossip is a round-based Synchronous Gossip Model which leads to high latency. To reduce the latency of data delivery, we circumvented the delay introducing steps of the Synchronous Model. We design a new gossip model called the Asynchronous Gossip Model which leads to faster and predictable data dissemination. Another key vi contribution is to analyze the behavior of gossip dissemination as a function of time instead of the conventional approach that uses fixed period rounds. To make the behavior of gossip protocols more predictable, we introduce a concept of adaptive fanout. Using the adaptive fanout, we can achieve fine-grained control of the rate at which gossip spreads a message to a group of nodes. Using our enhancements, we can make the dissemination of gossip messages closely follow user requirements, hence, predictable. We design adaptive fanout as a function of round for the Synchronous Gossip Model, and as a function of time for the Asynchronous Gossip Model. Through simulations, we show that the expected gossip behavior closely resembles our theoretical model. In the second part, we extend the work on Asynchronous Gossip to design a hierarchical gossip protocol which further increases the savings in number of gossip transmissions and reduces the latency of data delivery. More importantly, it improves the predictability of Asynchronous Gossip which is vital to the core of our research, i.e., making gossip more predictable. Organizing group nodes into a hierarchy or clusters based on performance criterion like latency or topological information is a widely studied approach to improve scalability and performance in distributed systems. We implement a hierarchical gossip protocol on a wide area network testbed (PlanetLab) and show that it outperforms the corresponding non-hierarchical flat global gossip protocol in terms of latency of data delivery. In particular, we implement Asynchronous Gossip on hierarchical gossip and show that the performance of Asynchronous Gossip is more predictable compared to the corresponding implementation on the global network. In our work, we use research ideas from network coordinates and the k-means clustering algorithm to design a centralized node clustering algorithm. Our results on node clustering demonstrates that using network coordinates is more efficient as well as reliable approach to distance based clustering instead of using direct measurements which lead to high processing overhead. We show improvements in transmission overhead, improved latency in data delivery and an improved predictability in the performance of Asynchronous Gossip. In the third part, we address that of high transmission overhead in gossip-based dissemination. Compared to tree-based deterministic protocol which require O(N ) transmissions to disseminate a message to a group of N nodes, push-based gossip needs O(N ln N ). This drawback makes push gossip very unattractive to designers. To alleviate this problem, we investigate the behavior of push gossip, and find that the message overhead in terms of message duplicates increases as the fraction of nodes that receive a gossip message increases. We use this observation to use push gossip to infect a random but fractional part of the entire group. To achieve successful gossip to all N nodes, we enhance partial push gossip with rateless codes to design Rateless Gossip. We vii show through analysis and simulations that Rateless Gossip indeed outperforms naive push gossip in terms of transmission overhead. Next, we further increase message savings in Rateless Gossip by pragmatic changes like using a hybrid membership mechanism and adding control messages. We call this the Optimized Rateless Gossip, and show that the average number of transmission required is O(cN ) where c can be fine-tuned based on gossip and coding parameters. viii Biographical Sketch Satish Verma was born on the 13th of September, 1978 in the city of Ballia in Uttar Pradesh, India. After he completed his secondary schooling at the D.A.V. Jawahar Vidya Mandir School in 1996, he went on to pursue his undergraduate degree in the Department of Electrical Engineering, at the Indian Institute of Technology, Madras. He graduated with a Bachelor Degree in Electrical Engineering in 2000. From 2000 to 2001, he attended EPFL where he earned a graduate degree in Communication Systems. In January 2003, he moved to Singapore to pursue a Ph.D. degree in School of Computing, National University of Singapore. ix Chapter Conclusions and Future Work In this dissertation, we have addressed two key challenges that handicap the performance of gossip protocols, namely, that of high and random latency of data delivery, and that of high transmission overhead. We design new analytical gossip models and protocols, i.e, the Asynchronous Gossip and the Rateless Gossip, which effectively address these drawbacks. Our research contributes to a deeper understanding of how gossip works and leads to improved performance of gossip protocols. There are, however, many other issues that warrant further research. In this chapter, we conclude our work and present possible extensions as future work. 6.1 Gossip Protocols with Predictable Behavior over Time We have designed gossip protocols which result in a more predictable behavior in terms of gossip infection spread rate over rounds/time. This is achieved by adapting fanout as a function of round/time. We show how the infection rate for the Synchronous Model can be made to follow a desired user defined infection pattern by defining fanout as a function of rounds. To reduce the latency of data delivery in Synchronous Model, we introduced the Asynchronous Gossip Model. We see that the spread of a message in Asynchronous Gossip is faster compared to Synchronous Gossip Model, and can be made to follow a user defined infection pattern over time by adapting fanout as a function of time. To fully comprehend the details of the Asynchronous Gossip, we introduce a hypothetical model called the PseudoSynchronous Gossip Model which provides a better understanding of gossip infection progress as a function of time. An important point to note here is that our analysis deals with expected values. It will be interesting to analytically quantify the variance and standard deviation of Asynchronous Gossip Performance and study 137 constraints on fanout values and time intervals that will lead to a more predictable performance. At the same time, how to design good user infection patterns is important since not all user patterns lead to solutions for the PseudoSynchronous Gossip Model parameters. Other problems worth studying are regarding efficient computation of the delay distribution whose knowledge we assume beforehand, and using partial membership views instead of global membership views. 6.2 Hierarchical Gossip We have presented a Hierarchical Gossip Protocol which improves the performance of a flat global gossip protocol by clustering nodes into groups based on inter-node latency criterion. We use techniques from well know research ideas like k-means clustering and virtual network coordinates. Significantly, we evaluate the performance of our Asynchronous Gossip Model on two networks. The first one is the global gossip which consists of one cluster with N nodes and the second one is the hierarchical gossip where the N nodes are clustered into multiple groups. We show that the performance of Asynchronous Gossip is better in terms of transmission overhead and latency of data delivery. More importantly, we show that the variance in the performance of Asynchronous Gossip is much smaller in case of hierarchical gossip when compared to global gossip. This proves that Asynchronous Gossip is quite predictable in small clusters with delay distribution spanning a small range of inter-node latencies. Our work can be extended in many way. We have used a centralized approach to cluster nodes and then send control messages to various nodes. A distributed clustering approach can be designed. Again, we compute all the delay distributions at the source node and then distribute the information to other nodes. This can be done at the cluster leaders since the delay distribution only involves the knowledge of inter-node latencies between cluster members. This improvement can make the hierarchical gossip protocol more scalable. Another possible extension would be to study the hierarchical gossip under more dynamic conditions and see the performance in terms of reliability, message overhead (duplicates plus control overhead) and latency. Finally, a gossip protocol combining Asynchronous Gossip with Rateless Gossip can be designed and studied for latency performance of data delivery using Rateless Gossip. 138 6.3 Rateless Gossip We have designed Rateless and Optimized Rateless Gossip Protocols which leads to substantial transmission savings in push gossip. We have used rateless codes in conjunction with gossip to design a low transmission-overhead push-based gossip protocol. We see from our simulation results that savings are in the range of 50 − 85%. We have provided a thorough analytical analysis of Rateless Gossip and Optimized Rateless Gossip which is backed up by experimental results. We showed that the average number of transmissions to gossip k messages in Optimized Rateless Gossip can be bounded as O(kcN ) where c is a tunable parameter. Recall that c depends on the gossip overhead factor α and the LT Code overhead factor . Future work should address how to minimize c for a given k and N . We have investigated the parameter α in details but the issue of choosing an optimal has not been considered. By understanding as to how to optimize , designers can achieve further savings. Another line of work could be to use more efficient rateless codes like the Raptor Code instead of LT Codes. We chose LT Codes due to availability of analytical results which helped us analyze Rateless Gossip in detail. However, an actual implementation can use the most optimal rateless code available such that is as low as possible to maximize savings. Theoretical analysis of Rateless Gossip for other rateless codes may be studied.We showed that by adapting the membership model and adding control messages, transmission overhead can be reduced in Rateless Gossip. More such optimizations like Anti-Entropy based pull-style approach can be used to optimize Rateless Gossip towards the later stages where a lot of useless messages are received. Furthermore, the performance of rateless codes for multicast/broadcast in a real network can be studied. In our work, we have a centralized way to generated encoded messages. It will be interesting to study ways to design Rateless Gossip in a distributed fashion when there are multiple sources in the network with each source having a fraction of interesting information. Our work focuses on improving the performance of gossip protocols in two key areas, i.e., to improve the predictability with respect to time, and, to reduce the transmission overhead incurred during large-scale data dissemination. Our analytical models of Asynchronous Gossip and Rateless Gossip can be combined and enhanced to design latency and message efficient gossip protocols for large-scale multicast applications. 139 Bibliography [1] Bittorrent. http://bitconjurer.org/bittorrent/protocol.html. [2] Comon, http://comon.cs.princeton.edu. [3] http://pyxida.sourceforge.net/. [4] Kazaa media dekstop. http://www.kazaa.com. [5] Planetlab. www.planet-lab.org. [6] I. Abraham, A. Badola, D. Bickson, D. Malkhi, S. Maloo, and S. Ron. Practical LocalityAwareness for Large Scale Information Sharing], booktitle = Proc. of IPTPS, year = 2005, address=Ithaca, NY, USA. [7] Abhishek Agrawal and Henri Casanova. Clustering Hosts in P2P and Global Computing Platforms. In CCGRID ’03: Proceedings of the 3st International Symposium on Cluster Computing and the Grid, Washington, DC, USA, 2003. IEEE Computer Society. [8] R. Ahlswede, Ning Cai, S. Y. R. Li, and R. W. Yeung. Network Information Flow. IEEE Transactions on Information Theory, 46(4):1204–1216, 2000. [9] Andr´e Allavena, Alan Demers, and John E. Hopcroft. Correctness of a Gossip-based Membership Protocol. In Proceedings of the ACM Symposium on Principles of Distributed Computing, pages 292–301, Las Vegas, NV, USA, 2005. [10] Lorenzo Alvisi, Jeroen Doumen, Rachid Guerraoui, Boris Koldehofe, Harry Li, Robbert van Renesse, and Gilles Tredan. How Robust are Gossip-based Communication Protocols? SIGOPS Operating Systems Review, 41(5):14–18, 2007. [11] N. T. J. Bailey. The Mathematical Theory of Infection Diseases. Griffin, 2nd Edition, 1975. 140 [12] Brenda Baker and Robert Shostak. Gossip and Telephones. Discrete Mathematics, pages 191–193, 1972. [13] Rena Bakhshi, Francois Bonnet, Wan Fokkink, and Boudewijn Haverkort. Formal Analysis Techniques for Gossiping Protocols. SIGOPS Operating Systems Review, 41(5):28–36, 2007. [14] S. Banerjee, B. Bhattacharjee, and C. Kommareddy. Scalable Application Layer Multicast. In ACM SIGCOMM, Pittsburgh, PA, USA, 2002. [15] S. Banerjee, S. Lee, B. Bhattacharjee, and A. Srinivasan. Resilient Multicast using Overlays. In ACM SIGMETRICS Performance Evaluation Review, pages 102–113, San Diego, CA, USA, 2003. [16] Roberto Beraldi. The Polarized Gossip Protocol for Path Discovery in MANETs. Ad Hoc Networks, 6(1):79–91, 2008. [17] A. Bestavros and S. Mehrotra. DNS-based Internet Client Clustering and Characterization. In WWC ’01: Proceedings of the Workload Characterization, 2001. WWC-4. 2001 IEEE International Workshop on, pages 159–168, Washington, DC, USA, 2001. IEEE Computer Society. [18] Ken Birman. The Promise, and Limitations, of Gossip Protocols. SIGOPS Operating Systems Review, 41(5):8–13, 2007. [19] Kenneth P. Birman, Mark Hayden, Oznur Ozkasap, Zhen Xiao, Mihai Budiu, and Yaron Minsky. Bimodal Multicast. ACM Transactions on Computer Systems, 17(2):41–88, 1999. [20] Stephen Boyd, Arpita Ghosh, Balaji Prabhakar, and Devavrat Shah. Randomized Gossip Algorithms. IEEE Transactions on Information Theory, 52(6):2508–2530, 2006. [21] David Braginsky and Deborah Estrin. Rumor Routing Algorthim for Sensor Networks. In Workshop on Wireless Sensor Networks and Applications, pages 22–31, Atlanta, GA, USA, 2002. [22] Mike Burmester, Tri Van Le, and Alec Yasinsac. Adaptive Gossip Protocols: Managing Security and Redundancy in Dense Ad hoc Networks. Ad Hoc Networks, 5(3):313–323, 2007. [23] J. Byers, M. Luby, and M. Mitzenmacher. A Digital Fountain Approach to Reliable Distribution of Bulk Data. In ACM SIGCOMM, pages 56–67, Vancouver, Canada, 1998. 141 [24] J. Byers, M. Luby, and M. Mitzenmacher. A Digital Fountain Approach to Asynchronous Reliable Multicast. IEEE Journal on Selected Areas in Communications, 20(8):1528–1540, 2002. [25] M. Castro, P. Druschel, A-M. Kermarrec, and A. Rowstron. SCRIBE: A Large-Scale and Decentralised Application-Level Multicast Infrastructure . IEEE Journal on Selected Areas in Communications, 20(8):379–390, 2002. [26] R. Chandra, V. Ramasubramanian, and K. Birman. Anonymous Gossip: Improving Multicast Reliability in Mobile Ad hoc Networks. In International Conference on Distributed Computing Systems, Phoenix, Arizona, USA, 2001. [27] Yang CHEN, Beixing DENG, and Xing LI. Experimental study on network coordinate based node clustering. Journal of Dalian University of Technology, 45(1):41–43, 2005. [28] Y. H. Chu, S. G. Rao, and H. Zhang. A Case for End System Multicast . In ACM SIGMETRICS, pages 1–12, Santa Clara, CA, USA, 2000. [29] Manuel Costa, Miguel Castro, Antony Rowstron, and Peter Key. PIC: Practical Internet Coordinates for Distance Estimation. In ICDCS ’04: Proceedings of the 24th International Conference on Distributed Computing Systems (ICDCS’04), pages 178–187, Washington, DC, USA, 2004. IEEE Computer Society. [30] Frank Dabek, Russ Cox, Frans Kaashoek, and Robert Morris. Vivaldi: A Decentralized Network Coordinate System. In SIGCOMM ’04: Proceedings of the 2004 conference on Applications, technologies, architectures, and protocols for computer communications, pages 15–26, New York, NY, USA, 2004. ACM. [31] A. Datta, S. Quarteroni, and K. Aberer. Autonomous Gossiping: A Self-organizing Epidemic Algorithm for Selective Information Dissemination in Mobile Ad hoc Networks. Lecture Nodes in Computer Science, 3226:126143, 2004. [32] Supratim Deb, Muriel Medard, and Clifford Choute. Algebraic Gossip: A Network Coding Approach to Optimal Multiple Rumor Mongering. IEEE/ACM Transactions on Networking, 14(SI):2486–2507, 2006. [33] A. Demers, D. Greene, C. Hauser, W. Irish, J. Larson, S. Shenker, H. Sturgis, D. Swinehart, and D. Terry. Epidemic Algorithms for Replicated Database Maintainance. In Proceedings 142 of ACM Symposium on Principles of Distributed Computing, Vancouver, British Columbia, Canada, 1987. [34] A. G. Dimakis, V. Prabhakaran, and K. Ramchandran. Decentralized Erasure Codes for Distributed Networked Storage. IEEE/ACM Transactions on Networking, 14:2809 – 2816, 2006. [35] Alexandros G. Dimakis, Anand D. Sarwate, and Martin J. Wainwright. Geographic Gossip: Efficient Aggregation for Sensor Networks. In Proceedings of the International Conference on Information Processing in Sensor Networks, pages 69–76, New York, NY, USA, 2006. [36] Max Val Machado, Raquel A.F. Mini, Antonio A. F. Loureiro, Daniel L. Guidoni, and Pedro O. S. V. de Melo. Gossiping Using the Energy Map in Wireless Sensor Networks. In Proceedings of the ACM Symposium on Modeling, Analysis, and Simulation of Wireless and Mobile Systems, pages 368–372, New York, NY, USA, 2007. [37] P. Th. Eugster, R. Guerraoui, S. B. Handurukande, P. Kouznetsov, and A.-M. Kermarrec. Lightweight Probabilistic Broadcast. ACM Transactions on Computer Systems, 21(4):341– 374, 2003. [38] Patrick Th. Eugster and Rachid Guerraoui. Probabilistic Multicast. In Proceedings of the International Conference on Dependable Systems and Networks, pages 313–324, Washington, DC, USA, 2002. [39] Y. Fernandess and D. Malkhi. On Collaborative Content Distribution Using Multi-Message Gossip. In 20th IPDPS, Rhodes Island, Greece, 2006. [40] Yaacov Fernandess, Antonio Fern´andez, and Maxime Monod. A Generic Theoretical Framework for Modeling Gossip-based Algorithms. SIGOPS Operating Systems Review, 41(5):19– 27, 2007. [41] Sally Floyd, Van Jacobson, Ching-Gung Liu, Steven McCanne, and Lixia Zhang. A Reliable Multicast Framework for Light-weight Sessions and Application Level Framing. IEEE/ACM Transactions on Networking, 5(6):784–803, 1997. [42] P. Francis, S. Jamin, C. Jin, Y. Jin, V. Paxson, D. Raz, Y. Shavitt, and L. Zhang. IDMaps: A Global Internet Host Distance Estimation Service. Networking, IEEE/ACM Transactions on, 9(5):525–540, 2001. 143 [43] Roy Friedman, Daniela Gavidia, Luis Rodrigues, Aline Carneiro Viana, and Spyros Voulgaris. Gossiping on MANETs: the Beauty and the Beast. SIGOPS Operating Systems Review, 41(5):67–74, 2007. [44] Ayalvadi J. Ganesh, Anne-Marie Kermarrec, and Laurent Massoulie. SCAMP: Peer-to-Peer Lightweight Membership Service for Large-Scale Group Communication. In Proceedings of the International COST264 Workshop on Networked Group Communication, pages 44–55, London, UK, 2001. [45] Ayalvadi J. Ganesh, Anne-Marie Kermarrec, and Laurent Massouli´e. HiScamp: Self- organizing Hierarchical Membership Protocol. In ACM SIGOPS European Workshop, pages 133–139, Saint-Emilion, France, 2002. [46] Ayalvadi J. Ganesh, Anne-Marie Kermarrec, Erwan Le Merrer, and Laurent Massouli´e. Peer Counting and Sampling in Overlay Networks Based on Random Walks. Distributed Computing, 20(4):267–278, 2007. [47] C. Gkantsidis and P.R. Rodriguez. Network Coding for Large-scale Content Distribution. In IEEE INFOCOM, pages 313–324, Miami, Fl, USA, 2005. [48] Krishna P. Gummadi, Stefan Saroiu, and Steven D. Gribble. King: estimating latency between arbitrary internet end hosts. In IMW ’02: Proceedings of the 2nd ACM SIGCOMM Workshop on Internet measurment, pages 5–18, New York, NY, USA, 2002. ACM. [49] Katherine Guo, Mark Hayden, Robbert van Renesse, Werner Vogels, and Kenneth P. Birman. GSGC: An Efficient Gossip-Style Garbage Collection Scheme for Scalable Reliable Multicast. Technical Report TR-97-1656, Cornell University, Ithaca, NY, USA, 1997. [50] I. Gupta, K. P. Birman, and R. van Renesse. Fighting Fire with Fire: Using Randomized Gossip To Combat Stochastic Scalability Limits. Quality and Reliability Engineering International, 18:165–184, 2002. [51] Indranil Gupta, Ken Birman, Prakash Linga, Al Demers, and Robbert van Renesse. Kelips: Building an Efficient and Stable P2P DHT Through Increased Memory and Background Overhead. In Proceedings of the International Workshop on Peer-to-Peer Systems (IPTPS), Berkeley, CA, USA, 2003. 144 [52] Indranil Gupta, Anne-Marie Kermarrec, and Ayalvadi J. Ganesh. Efficient and Adaptive Epidemic-Style Protocols for Reliable and Scalable Multicast. IEEE Transactions on Parallel Distributed Systems, 17(7):593–605, 2006. [53] Zygmunt J. Haas, Joseph Y. Halpern, and Li Li. Gossip-based Ad hoc Routing. IEEE/ACM Transactions on Networking, 14(3):479–491, 2006. [54] L. J. Heyer, S. Kruglyak, and S. Yooseph. Exploring Expression Data: Identification and Analysis of Coexpressed Genes. Genome Research, 9:1106–1115, 1999. [55] Hugh W. Holbrook, Sandeep K. Singhal, and David R. Cheriton. Log-Based Receiver-Reliable Multicast for Distributed Interactive Simulation. In ACM SIGCOMM, pages 328–341, Cambridge, Massachusetts, USA, 1995. [56] A. K. Jain, M. N. Murty, and P. J. Flynn. Data Clustering: A Review. ACM Computing Surveys, 31(3):264–323, 1999. ¨ [57] M´ark Jelasity and Ozalp Babaoglu. T-Man: Gossip-Based Overlay Topology Management. In Engineering Self-Organising Systems: Third International Workshop, ESOA, pages 1–15, Utrecht, The Netherlands, 2005. [58] Mark Jelasity and Anne-Marie Kermarrec. Ordered Slicing of Very Large-Scale Overlay Networks. In Proceedings of the IEEE International Conference on Peer-to-Peer Computing, pages 117–124, Washington, DC, USA, 2006. ¨ [59] M´ark Jelasity, Alberto Montresor, and Ozalp Babaoglu. Gossip-based Aggregation in Large Dynamic Networks. ACM Transactions on Computer Systems, 23(3):219–252, 2005. ¨ [60] Gian Paolo Jesi, Alberto Montresor, and Ozalp Babaoglu. Proximity-Aware Superpeer Overlay Topologies. IEEE Transactions on Network and Service Management, 4(2):43–57, 2006. [61] S. C. Johnson. Hierarchical Clustering Schemes . Psychometrika, 32(3):241–254, 1967. [62] Abhinav Kamra, Vishal Misra, Jon Feldman, and Dan Rubenstein. Growth Codes: Maximizing Sensor Network Data Persistence. In SIGCOMM ’06, pages 255–266, New York, NY, USA, 2006. ACM. [63] R. Karp, M. Luby, and A. Shokrollahi. Finite length analysis of LT codes. In Intl. Symposium on Information Theory (ISIT), page 39, Chicago, IL ,USA, 2004. 145 [64] Richard M. Karp, Christian Schindelhauer, Scott Shenker, and Berthold Vocking. Randomized Rumor Spreading. In IEEE Symposium on Foundations of Computer Science, pages 565–574, Redondo Beach, CA , USA, 2000. [65] S. Katti, H. Rahul, W. Hu, D. Katabi, M M. Medard, and J. Crowcroft. XORs in the Air: Practical Wireless Network Coding. In ACM SIGCOMM, pages 243–254, Pisa, Italy, 2006. [66] David Kempe, Alin Dobra, and Johannes Gehrke. Gossip-based Computation of Aggregate Information. In Proceedings of the IEEE Symposium on Foundations of Computer Science, pages 482–491, Washington, DC, USA, 2003. [67] David Kempe and Jon M. Kleinberg. Protocols and Impossibility Results for Gossip-Based Communication Mechanisms. In Proceedings of the Symposium on Foundations of Computer Science, pages 471–480, Washington, DC, USA, 2002. [68] David Kempe, Jon M. Kleinberg, and Alan J. Demers. Spatial Gossip and Resource Location Protocols. In ACM Symposium on Theory of Computing, pages 163–172, Heraklion, Crete, Greece, 2001. [69] Anne-Marie Kermarrec, Laurent Massouli´e, and Ayalvadi J. Ganesh. Probabilistic Reliable Dissemination in Large-Scale Systems. IEEE Transactions on Parallel Distributed Systems, 14(3):248–258, 2003. [70] B. Koldehofe. Simple Gossiping with Balls and Bins. In 6th International Conference on Principles of Distributed Systems (OPODIS), pages 109–118, Reims, France, 2002. [71] D. Kostic, A. Rodriguez, J. Albrecht, and A. Vahdat. Bullet: High Bandwidth Data Dissemination Using an Overlay Mesh. In Proc. of 19th ACM SOSP, pages 19–22, New York, USA, 2003. [72] Dionysios Kostoulas, Dimitrios Psaltoulis, Indranil Gupta, Kenneth P. Birman, and Alan J. Demers. Active and Passive Techniques for Group Size Estimation in Large-scale and Dynamic Distributed Systems. Journal of System Software, 80(10):1639–1658, 2007. [73] Balachander Krishnamurthy and Jia Wang. On Network-aware clustering of Web clients. In SIGCOMM ’00: Proceedings of the conference on Applications, Technologies, Architectures, and Protocols for Computer Communication, pages 97–110, New York, NY, USA, 2000. ACM. 146 [74] Rajnish Kumar, Arnab Paul, and Umakishore Ramachandran. Fountain Broadcast for Wireless Networks. Technical Report GIT-CERCS-05-02, Georgia Tech., GA, USA, 2005. [75] P. Kyasanur, R. R. Choudhury, and I. Gupta. Smart Gossip: An Adaptive Gossip-based Broadcasting Service for Sensor Networks. In IEEE International Conference on Mobile Ad hoc and Sensor Systems (MASS), pages 91–100, Vancouver, Canada, 2006. [76] Yan Li, Li Lao, and Jun-Hong Cui. SDC: A Distributed Clustering Protocol for Peer-to-Peer Networks. In Networking, pages 1234–1239, 2006. [77] Meng-Jang Lin and Keith Marzullo. Directional Gossip: Gossip in a Wide Area Network. In European Dependable Computing Conference, pages 364–379, Prague, Czech Republic, 1999. [78] Eng Keong Lua, Jon Crowcroft, and Marcelo Pias. Highways: Proximity Clustering for Scalable Peer-to-Peer Network. In Peer to Peer Computing, Zurich, Switzerland, 2004. [79] Michael Luby. LT Codes. In The Annual IEEE Symposium on Foundations of Computer Science, pages 271–282, Vancouver, BC, Canada, Nov 2002. [80] J. Luo, P. T. Eugster, and J. P. Hubaux. Route-driven Gossip: Probabilistic Reliable Multicast in Ad hoc Networks. In The Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM), pages 2229–2239, San Fransisco, CA, USA, 2003. [81] D. Mackay. Fountain Codes. IEE Proceedings on Communications, 152-6:1062–1068, 2005. [82] David J.C. MacKay. Information Theory, Inference, and Learning Algorithms. Cambridge University Press, 2003. [83] Dahlia Malkhi, Elan Pavlov, and Yaron Sella. Gossiping with Malicious Parties. Technical Report TR-2003-09, School of Computer Science and Engineering, The Hebrew University of Jerusalem, Israel, 2003. [84] Balasubramaneyam Maniymaran, Marin Bertier, and Anne-Marie Kermarrec. Build One, Get One Free: Leveraging the Coexistence of Multiple P2P Overlay Networks. In IEEE International Conference on Distributed Computing Systems (ICDCS), Toronto, Ontario, Canada, 2007. [85] L. Mathy, R. Canonico, S. Simpson, and D. Hutchison. Scalable Adaptive Hierarchical Clustering . Communications Letters, IEEE, 6(3):117–119, 2002. 147 [86] P. Maymounkov. Online Codes. Technical Report TR-2003-883, New York University, NY, USA, 2002. [87] P. Maymounkov and D. Mazieres. Rateless Codes and Big Downloads. In Proceedings of the International Workshop on Peer-to-Peer Systems (IPTPS), Berkeley, CA, USA, 2003. [88] P. Maymounkov and D. Mazieres. Rateless Codes and Big Downloads. In 2nd IPTPS, pages 1190– 1199, Berkeley, CA, USA, 2003. [89] Yaron M. Minsky and Fred B. Schneider. Tolerating Malicious Gossip. Distributed Computing, 16(1):49–68, 2003. [90] M. Mitzenmacher and E. Upfal. Probability and Computing. Cambridge University Press, 2005. [91] C. C. Moallemi and B. Van Roy. Consensus Propagation. IEEE Transactions on Information Theory, 52(11):4753–4766, 2006. [92] E. Ng and H. Zhang. Predicting Internet Network Distance with Coordiantes-based Approaches. In IEEE Conference on Computer Communications (INFOCOM ’01), Anchorage, Alaska, USA, 2001. [93] M.K. Ng. A Note on Constrained k-means Algorithms. Pattern Recognition, 33(3):515–519, March 2000. [94] T. S. E. Ng and Hui Zhang. Predicting Internet Network Distance with Coordinates-based Approaches. In INFOCOM 2002. Twenty-First Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings. IEEE, pages 170–179, 2002. [95] Ning Ning, Dongsheng Wang, Yongquan Ma, Jinfeng Hu, Jing Sun, Chongnan Gao, and Weimin Zheng. Genius: Peer-to-Peer Location-Aware Gossip Using Network Coordinates. In International Conference on Computational Science, pages 163–170, Atlanta, U.S.A, 2005. [96] G. Pandurangan. Building Low-Diameter P2P Networks. Proceedings of the IEEE symposium on Foundations of Computer Science, 21(6):995–1002, 2001. [97] J.A. Patel, I. Gupta, and N. Contractor. JetStream: Achieving Predictable Gossip Dissemination by Leveraging Social Network Principles. In International Symp. on Network and Computing Applications (NCA), pages 32–39, Cambridge, MA, USA, 2006. 148 [98] Sanjoy Paul, Krishan K. Sabnani, John C.-H. Lin, and Supratik Bhattacharyya. Reliable Multicast Transport Protocol (RMTP). IEEE Journal of Selected Areas in Communications, 15(3):407–421, 1997. [99] Marcelo Pias, Jon Crowcroft, Steve Wilbur, Tim Harris, and Saleem Bhatti1. Lighthouses for Scalable Distributed Location. Lecture Notes in Computer Science, 2735/2003:278–291, 2003. [100] Peter Pietzuch, Jonathan Ledlie, Michael Mitzenmacher, and Margo Seltzer. Network-Aware Overlays with Network Coordinates. In Proceedings of IWDDS 2006, Lisbon, Portugal, July 2006. [101] Peter Pietzuch, Jonathan Ledlie, and Margo Seltzer. Supporting network coordinates on planetlab. In Second Workshop on Real, Large Distributed Systems, San Francisco, CA, December 2005. [102] Nazanin Rahnavard and Faramarz Fekri. CRBcast: A Collaborative Rateless Scheme for Reliable and Energy-efficient Broadcasting in Wireless Sensor Networks. In ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN), pages 276– 283, Nashville, Tennessee, USA, 2006. [103] S. Ratnasamy, M. Handley, R. Karp, and S. Shenker. Application-Level Multicast Using Content-Addressable Networks. Lecture Notes in Computer Science, 2233:161–172, 2001. [104] S. Ratnasamy, M. Handley, R. Karp, and S. Shenker. Topologically-Aware Overlay Construction and Server Selection. In IEEE INFOCOM, pages 1190– 1199, New York, USA, 2002. [105] Robbert Van Renesse, Kenneth P. Birman, and Werner Vogels. Astrolabe: A Robust and Scalable Technology for Distributed System Monitoring, Management, and Data Mining. ACM Transactions on Computer Systems, 21(2):164–206, 2003. [106] L. Rodrigues, S. Handurukande, J. Pereira, R. Guerraoui, and A. Kermarrec. Adaptive Gossip-based Broadcast. In Dependable Systems and Networks, pages 47 – 56, San Francisco, California, USA, 2003. [107] Antony I. T. Rowstron and Peter Druschel. Pastry: Scalable, Decentralized Object Location, and Routing for Large-Scale Peer-to-Peer Systems. In Proceedings of the IFIP/ACM 149 International Conference on Distributed Systems Platforms Heidelberg (Middleware), pages 329–350, London, UK, 2001. [108] Rik Sarkar, Xianjin Zhu, and Jie Gao. Hierarchical Spatial Gossip for Multi-resolution Representations in Sensor Networks. In Proceedings of the International Conference on Information Processing in Sensor Setworks, pages 420–429, New York, NY, USA, 2007. [109] Y. Sasson, D. Cavin, and A. Schiper. Probabilistic Broadcast for Flooding in Wireless mobile Ad hoc Networks. In Proceedings of IEEE Wireless Communications and Networking Conference (WCNC), New Orleans, Louisiana, USA, 2003. [110] S.Bandyopadhyay and E. J. Coyle. An Energy Efficient Hierarchical Clustering Algorithm for Wireless Sensor Networks . In INFOCOM 2003. Twenty-Second Annual Joint Conference of the IEEE Computer and Communications Societies. IEEE, pages 1713– 1723, San Fransisco, CA, USA, 2003. [111] Yuval Shavitt and Tomer Tankel. Big-bang simulation for embedding network distances in euclidean space. IEEE/ACM Trans. Netw., 12(6):993–1006, 2004. [112] A. Shokrollahi. Raptor Codes. IEEE/ACM Transactions on Networking, 14:2551–2567, 2006. [113] Ion Stoica, Robert Morris, David Karger, M. Frans Kaashoek, and Hari Balakrishnan. Chord: A Scalable Peer-to-peer Lookup Service for Internet Applications. In ACM SIGCOMM, pages 149–160, San Diego, CA,USA, 2001. [114] Qixiang Sun and Daniel C. Sturman. A Gossip-Based Reliable Multicast for Large-Scale High-Throughput Applications. In Proceedings of the ICDSN, New York, NY, USA, 2000. [115] Stijn van Dongen. A New Cluster Algorithm for Graphs. Technical Report INS-R9814, Centrum voor Wiskunde en Informatica (CWI), 1998. [116] R. van Renesse, Y. Minsky, and M. Hayden. A Gossip-style Failure-detection Service. In Proceedings of IFIP International Conference on Distributed Systems Platforms and Open Distributed Processing, pages 55–70, The Lake District, England, 1998. [117] Satish Verma and Wei Tsang Ooi. Controlling Gossip Protocol Infection Pattern Using Adaptive Fanout. In IEEE International Conference on Distributed Computing Systems (ICDCS), 2005. 150 [118] Spyros Voulgaris, Daniela Gavidia, and Maarten v Steen. CYCLON: Inexpensive Membership Management for Unstructured P2P Overlays. Journal of Network and Systems Management, 13(2):197–217, 2005. [119] Spyros Voulgaris, Anne-Marie Kermarrec, Laurent Massoulie, and Maarten v Steen. Exploiting Semantic Proximity in Peer-to-Peer Content Searching. In Proceedings of the IEEE International Workshop on Future Trends of Distributed Computing Systems, pages 238–243, Washington, DC, USA, 2004. [120] Spyros Voulgaris, Maarten van Steen, and Konrad Iwanicki. Proactive Gossip-based Management of Semantic Overlay Networks: Research Articles. Concurrency and Computation: Practice and Experience, 19(17):2299–2311, 2007. [121] M. Waldvogel and R. Rinaldi. Efficient Topology-Aware Overlay Network. SIGCOMM Rev., 33(1), 2003. [122] Hakim Weatherspoon, Hugo Miranda, Konrad Iwanicki, Ali Ghodsi, and Yann Busnel. Gossiping Over Storage Systems is Practical. SIGOPS Operating Systems Review, 41(5):75–81, 2007. [123] R. Wittmann and M. Zitterbart. Multicast Communication: Protocols, Programming, and Applications. Morgan Kaufmann Publishers, May, 2000. [124] Bernard Wong, Aleksandrs Slivkins, and Emin Gn Sirer. Meridian: A Lightweight Network Location Service without Virtual Coordinates. In Proceedings of SIGCOMM Conference, August 2005, 2005. [125] Chuan Wu and Baochun Li. rStream: Resilient Peer-to-Peer Streaming with Rateless Codes. In ACM MULTIMEDIA, pages 307–310, New York, NY, USA, 2005. [126] Zhen Xiao and Kenneth P. Birman. A Randomized Error Recovery Algorithm for Reliable Multicast. In Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM), pages 239–248, Anchorage, Alaska, USA, 2001. [127] Meng Zhang, Li Zhao, Yun Tang, Jian-Guang Luo, and Shi-Qiang Yang. Large-scale Live Media Streaming Over Peer-to-Peer Networks Through Global Internet. In Proceedings of the ACM workshop on Advances in Peer-to-Peer Multimedia Streaming, New York, NY, USA, 2005. 151 [128] Xinyan Zhang, Jiangchuan Liu, Bo Li, and Tak-Shing Peter Yum. CoolStreaming/DONet: A Data-Driven Overlay Network for Efficient Live Media Streaming. In Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM), Miami, Fl, USA, 2005. [129] B. Zhao, L. Huang, J. Stribling, S. Rhea, A. Joseph, and J. Kubiatowicz. Tapestry: A Resilient Global-scale Overlay for Service Deployment. IEEE Journal on Selected Areas in Communications, 22(1), 2003. [130] J. Zhao, F. Yang, Q. Zhang, Z. Zhang, and F. Zhang. LION: Layered Overlay Multicast with Network Coding. IEEE Trans. on Multimedia, 8(5):1021–1032, 2006. [131] S. Zhong and D. DeMaris. Diameter-constrained Clustering: Algorithms and Experiments for a Layout Coverage Problem. In 7th Annual Austin CAS International Conference, Austin, USA, 2006. 152 [...]... participation of all the peers A large number of analytical results can on found on topics such as aggregation protocols, efficient overlay construction, and for ad hoc and sensor network protocols which are designed using gossip P2P networks and sensor networks are flourishing technologies and their nature solicits more and more usage of gossip And to understand the performance bounds and limitations of gossip for. .. Analysis of Push Gossip Protocol: Im vs m Analysis of Push Gossip Protocol: E[Mk ] and V ar[Mk ] Expected Value,E[i|0 i] and Variance, V ar[i|0 i] m m Distribution of 0 i for different m m Example of gossip progress for h-hop Gossip Protocol h Distribution 0 i for h-hop Gossip Protocol h Comparison between 0 i and 0 i 5.8 5.9 5.10 5.11 5.12 5.13 Message Distribution... responsible for group management, routing and data forwarding 1.3.3 Randomized Gossip Protocols Another approach to large-scale multicast applications is to use randomized gossip or epidemic protocols In this case, the group members keep a partial overview of the group in form of a membership view From this membership view, nodes are picked randomly and data is forwarded to them The key advantage of such... a gossip message will be Therefore, to control the rate of message dissemination and thus the latency, it is essential to control and adapt the gossip fanout We present and analyze two models for gossip- based data dissemination, namely, the round-based Synchronous Gossip Model and the time-based Asynchronous Gossip Model For both the models, we show analytically and experimentally how the gossip process... Distribution: Gossip splits the total load over all the processes symmetrically Topology Independence: Gossip performs well for most reasonably well-connected topologies Ease of Local Information Discovery: Gossip scales in discovery nearby resources better than flooding Table 2.2: Weaknesses of Gossip Protocols Limitations of Gossip Protocols Slow: Gossip is inherently slow due to periodic message exchange and. .. not spreading the information anymore These process states are extremely useful for the mathematical modeling of the epidemic process when one wants to keep track of the progress of the gossip protocol in terms of the number of infected, susceptible and removed processes The notion also help one measure the performance of gossip protocols For example, one can estimate the number of gossip steps, i.e.,... according to the distance of the process from the source There are various models for gossip- based protocols to measure the performance based on the distance metric, like Uniform and Spatial gossip models [68] In Uniform Gossip, in each gossip step, each node u chooses a node v randomly and uniformly from the entire network and updates it This is very similar to the Flat Gossiping scheme we discussed... Asynchronous Gossip Performance of Global Gossip Delay Pdfs for Hierarchical Gossip, Cluster 1 Adaptive Fanout for Hierarchical Gossip, Cluster 1 Asynchronous Gossip Performance of Hierarchical Gossip, Cluster 1 Delay PDFs for Hierarchical Gossip, Cluster 2 Adaptive Fanout for Hierarchical Gossip, Cluster 2 ... they have already received the message, leading to a larger number of hops before all nodes receive a copy Not only is the latency of delivery high, but it is also random since subsequent gossip messages take different paths These problems make gossip an unsuitable choice for soft-real time applications Thus, tackling the high latency of message delivery and making gossip performance more predictable with... of time can be an interesting approach to study the progress and properties of gossip- based protocols Another interesting point worth mentioning about the theoretical properties of gossip is the approach used in analyzing their performance The performance metric could be a measure of the number of rounds, number of messages transmitted etc Thus, complexity analysis of algorithms finds quite a lot of . 139 v Abstract Techniques for Improving Predictability and Message Efficiency of Gossip Protocols Satish Kumar Verma National University of Singapore Gossip- based protocols are a class of randomized. Techniques for Improving Predictability and Message Efficiency of Gossip Protocols SATISH KUMAR VERMA B.Tech., IIT Madras A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY SCHOOL OF. by gossip algorithms, and propose techniques to improve the efficiency and performance of gossip protocols. The first challenge we tackle is the high and random latency of data delivery that gossip protocols