Welcome to the ADMT Publication Server

The Multicast Pull Advantage in Dissemination-based Data Delivery

DocUID: 2004-009 Full Text: PDF

Author: Jonathan Beaver, Kirk Pruhs, Panos K. Chrysanthis, Vincenzo Liberatore

Abstract: A major problem in web database applications and on the Internet in general is the scalable delivery of data. Multicast is one of the standard techniques to achieve scalable data dissemination. However the use of multicast introduces a variety of data management issues at the server. In this paper we examine three major problems namely, the push popularity problem, the document classification problem and the bandwidth division problem, that arise in the design of a hybrid data dissemination scheme. We propose solutions to these problems and argue that these are essentially the best possible solutions. In particular, we argue for having a multicast pull mode, in addition to the traditional unicast pull mode and the commonly proposed multicast push mode. We give a simple method for estimating the current popularities of pushed documents. We give an algorithm for determining which documents should be pushed/pulled, and for determining how much of the server bandwidth should be devoted to push/pull. We report on experiments with our system that validate our algorithms.

Published In: 3rd Hellenic Data Management Symposium

Place Published: Athens, Greece

Year Published: 2004

Project: Others Subject Area: Data Dissemination, Database Servers

Publication Type: Conference Paper

Sponsor: NSF ANI-0123705

Citation:Text Latex BibTex XML Jonathan Beaver, Kirk Pruhs, Panos K. Chrysanthis, and Vincenzo Liberatore. The Multicast Pull Advantage in Dissemination-based Data Delivery, 3rd Hellenic Data Management Symposium (HDMS'04), Athens, Greece, June 2004.