Barboni, S. (2011):
Assessment of dependability scenarios in large-scale grids, using GridSim toolkit
Grid Computing refers to coordinated resource sharing and problem solving in dynamic multi-institutional virtual organizations (VOs). As grid users (i.e., the members of the vir- tual organizations) are basing their work more and more on grid technology, grid systems must exhibit a high degree of dependability, i.e., they must be able to deliver service that can justiably be trusted. Although there have been several research efforts established to address dependability issues in distributed systems, most of the characteristics inherent to grids are considered only recently. One such characteristic is the large-scaleness of grids which presents a major challenge in understanding dependability. In this thesis we simulate large-scale grids in order to enable a systematic study of their inherent dynamics (VOs, applications, middleware, resources, and networks). The goal is to obtain the scenarios where we can get the description of the behavior of its components on a large scale. In this thesis we propose the use of modeling and simulation, because various grid scenarios need to be evaluated and repeated. Hence, this thesis describes the development of GridSim, a discrete-event grid simulation tool, which allows modeling and simulation of various properties. The simulation is based on the GridSim tools provided by the University of Melbourne. The tool provide a virtual grid infrastructure that enables experimentation with dynamic resource management techniques and adaptive services by supporting controllable, repeatable, observable experiments.