Performance Improvement of Information Service Using Priority Driven Method

Abstract

Grid is developed to accomplish large and complex computation by gathering distributed computing resources. Grid employs information service to manage and provide these collected resources. The accurate information on resource provider is essential to stable service. In Grid information service, GRIS and GIIS are used to gather and maintain the resource information. These two directory servers store resource information in a cache only for a fixed period to improve fast response and accurate access. Since resource information search is performed in the cache, the system performance depends on the search time in a cache. In this paper, we propose a novel cache management system based on the resource priority. Priority of resource is determined by the frequency of resource usage, and the number of available resources in GRIS. The simulated priority driven schemes provide more accurate information and faster response to a client than the existing cache update scheme of information service.