Scheduling Strategies Of Correlated Broadcast Data In Push-Based Systems

Abstract

The server strategy of repeatedly broadcasting data items can result in a higher throughput than sending the requested data items to individual clients. Various methods have been studied to reduce the average response time in such systems. In this paper, two strategies axe presented for scheduling the broadcast program. Both strategies take into account the correlation among the requested data items. The first strategy is suitable for an environment where clients access a set of correlated data items by submitting multiple access requests at the same time. A broadcast program is composed by placing strongly correlated data items next to each other. The second strategy is suitable for an environment where clients access a set of correlated data items by submitting multiple 61CC6SS requests at intervals between the requests. This latter strategy places strongly correlated data items at suitable intervals in the program. Simulation studies were used to evaluate the performance of the proposed strategies. RESUMELa strategie qui consiste pour un serveur a diffuser de maniere repetee des donnees, peut conduire a un meilleur debit que d'envoyer les donnees a la demande de chaque client. DiflFerentes methodes ont ete etudiees afin de reduire le temps de reponse moyen dans de tels systemes. Dans cet article, nous presentons deux strategies pour l'ordonnancement du programme de diflEusion. Ces deux strategies prennent en compte les correlations entre les donnees demandees. La premiere strategie convient pour un environnement dans lequel les clients accedent a un ensemble de donnees correlees en soumettant de multiples demandes d'acces en meme temps. Un programme de diffusion est orgarnise en mettant les donnees fortement correlees l'une a cote de l'autre. La deuxieme strategie convient pour un environnement dans lequel les clients accedent a un ensemble de donnees fortement correlees en soumettant de multiples demandes d'acces par intervalles. Cette derniere strategie place les donnees fortement correlees a intervalles appropries dans le programme. Des etudes de simulation ont permis d'evaluer les performances des strategies proposees.

Query processing methods considering the deadline of queries for database broadcasting systems

“…There are many studies for improving the performance of broadcast information systems, which include scheduling techniques at the server side [1,4,7,[10][11][12], caching techniques at the client side [1], update propagation techniques [2], integration of push-based broadcast and pull-based broadcast [3,6], and pre-fetching techniques [5]. Most of them deal with broadcast data as data items simply, and do not address the performance improvement by considering contents and characteristics of broadcast data.…”