Choosing a load balancing scheme for agent-based digital libraries

Abstract

Digital Libraries (DLs) provide an important application area for deploying mobile agents, and with the increase in content being made available within such libraries, performance concerns are becoming significant. Current DLs often involve content servers which are geographically distributed, often necessitating information from these servers to be aggregated to answer a single query. Encoding a query as a mobile agent provides a useful means to implement such queries. However, a significant concern when this approach is adopted is the need to load balance multiple sets of such agents across the available servers. This paper focuses on an attempt to answer which load balancing scheme should be applied to an agent-based DL. A demonstration of a load balancing scheme based on the guidelines proposed in this paper with reference to Synthetic Aperture Radar Atlas (SARA) DL, consisting of multispectral images of the Earth, is presented. This particular DL involves processing image content, but also involves aggregation of data acquired from the images with text-based content available elsewhere.