GRID resource searching on the gridsim simulator

Abstract

Nowadays, the Grid is the focus of multiple researches. Our work is centered on Resource Management for Grids as it is an opened and current research area. Decentralized, scalable and efficient resource search algorithms are one of the key issues for resource management in large Grid systems. Resource search is required in order to allocate applications and data efficiently and to maintain the quality of service at runtime, just to mention some examples. In this work, we propose a scheme that presents essential characteristics for self-configuring search and is able to handle dynamic resources, such as memory capacity. Our approach consists on a hypercube topology connecting the nodes and a scalable and self-configuring search procedure. The algorithm improves the probability of reaching the alive nodes in the Grid even in the presence of non-alive ones (inaccessible, crashed or heavy loaded nodes). In this paper, after the theory's description, we present some results obtained by running our search protocol on the GridSim simulator. We have evaluated 6 different metrics performing several resources searches and we show the arithmetic media for each measure. © 2009 Springer Berlin Heidelberg.