Author : Vasil Georgiev Slavov
Publisher :
ISBN 13 :
Total Pages : 81 pages
Book Rating : 4.:/5 (815 download)
Book Synopsis A Study of Gossip Algorithms for Internet-scale Cardinality Estimation of Distributed XML Data by : Vasil Georgiev Slavov
Download or read book A Study of Gossip Algorithms for Internet-scale Cardinality Estimation of Distributed XML Data written by Vasil Georgiev Slavov and published by . This book was released on 2012 with total page 81 pages. Available in PDF, EPUB and Kindle. Book excerpt: After more than a decade of active research and development, the peer-to-peer (P2P) computing model continues to be successful. We have witnessed the deployment of commercial P2P applications in large, Internet-scale environments. With the rise and growth of P2P, indexing and querying data stored in large-scale sharing systems has become increasingly di cult. Computing statistics over data stored in Internet-scale P2P systems is an important component of query optimization. Decentralized gossip-based protocols are very popular in networking, and in particular, in sensor networks. The simplicity and scalability of gossip protocols render them perfect for quickly computing accurate estimates of aggregates (sums, averages, etc.) in Internet-scale systems where node and link failures are the norm. In this thesis, we present the problem of cardinality estimation of XPath queries over XML data stored in a distributed, Internet-scale environment. We focus our work on three objectives: implementing gossip in an Internet-scale environment, conducting a comprehensive performance evaluation in a wide-area network, and analyzing the experimental results. We implement two gossip-based algorithms (VanillaXGossip and XGossip) which, given an XPath query, estimate the number of XML documents in the network that contain a match for the query. XGossip employs a new, divide-and-conquer strategy for load-balancing and reducing the bandwidth consumption. We conduct a comprehensive performance evaluation of both gossip algorithms on Amazon Elastic Compute Cloud (Amazon EC2) web service using a heterogeneous collection of XML documents. The goal of the performance evaluation is to nd if the results we obtain are consistent with the theoretical analysis of VanillaXGossip and XGossip.