An increased need for collaborative research among different organizations, together with continuing advances in communication technology and computer hardware, has facilitated the development of distributed systems that can provide users non-trivial access to geographically dispersed computing resources (processors, storage, applications, data, instruments, etc.) that are administered in multiple computer domains. The term grid computing or grids is popularly used to refer to such distributed systems. A broader definition of grid computing includes the use of computing resources within an organization for running organization-specific applications. This research is in the context of using grid computing within an enterprise to maximize the use of available hardware and software resources for processing enterprise applications. Large scale scientific simulations have traditionally been the primary benefactor of grid computing. The application of this technology to simulation in industry has, however, been negligible. This research investigates how grid technology can be effectively exploited by simulation practitioners using Windows-based commercially available simulation packages to model simulations in industry. These packages are commonly referred to as Commercial Off-The-Shelf (COTS) Simulation Packages (CSPs). The study identifies several higher level grid services that could be potentially used to support the practise of simulation in industry. It proposes a grid computing framework to investigate these services in the context of CSP-based simulations. This framework is called the CSP-Grid Computing (CSP-GC) Framework. Each identified higher level grid service in this framework is referred to as a CSP-specific service. A total of six case studies are presented to experimentally evaluate how grid computing technologies can be used together with unmodified simulation packages to support some of the CSP-specific services. The contribution of this thesis is the CSP-GC framework that identifies how simulation practise in industry may benefit from the use of grid technology. A further contribution is the recognition of specific grid computing software (grid middleware) that can possibly be used together with existing CSPs to provide grid support. With its focus on end-users and end-user tools, it is intended that this research will encourage wider adoption of grid computing in the workplace and that simulation users will derive benefit from using this technology.

An emerging trend in computing is to use distributed heterogeneous computing (HC) systems to execute a set of tasks. Cluster computer systems, grids, and Desktop Grids are three popular kinds of HC systems. An important component of an HC system is its resource management system (RMS). The main responsibility of an RMS is assigning resources to tasks in order to satisfy certain performance requirements. For cluster computer systems, we propose a new mapping heuristic which requires less state information than current heuristics. For Desktop Grids, we propose a new scheduling policy that exploits knowledge of the effective computing power delivered by the machines and the distribution of their fault times in order to improve performance. Finally, for grids, we propose a new decentralized load balancing policy which dramatically cuts down the communication overhead incurred in state information update. The proposed resource management policies utilize the solution to a linear programming problem (LP) which maximizes the system capacity. Our simulation experiments show that these policies perform very competitively, especially in highly heterogeneous systems. / Thesis / Doctor of Philosophy (PhD)

This thesis deals with the mapping and the scheduling of workflows. In this context, we consider unreliable platforms, with processors subject to failures. In a first part, we consider a particular model of streaming applications : the filtering services. In this context, we aim at minimizing period and latency. We first neglect communication costs. In this model, we study scheduling problems on homogeneous and heterogeneous platforms. Then, the impact of communication costs on scheduling problems of a filtering application is studied. Finally, we consider the scheduling problem of such an application on a chain of processors. The theoretical complexity of any variant of this problem is proved. This filtering property can model the reliability of processors. The results of some computations are successfully computed, and some other ones are lost. We consider the more frequent failure types : transient failures. We aim efficient and reliable schedules. The complexity of many variants of this problem is proved. Two heuristics are proposed and compared using using simulations. Even if transient failures are the most common failures in classical grids, some particular type of platform are more concerned by other type of problems. Desktop grids are especially unstable. In this context, we want to execute iterative applications. All tasks are executed, then a synchronization occurs, and so on. Two variants of this problem are considered : applicationsof independent tasks, and applications where all tasks need to be executed at same speed. In both cases, the problem is first theoretically studied, then heuristics are proposed and compared using simulations.

The execution of parallel discrete event simulation across metacomputing infrastructures is examined. A master/worker architecture for parallel discrete event simulation is proposed providing robust executions under a dynamic set of services with system-level support for fault tolerance, semi-automated client-directed load balancing, portability across heterogeneous machines, and the ability to run codes on idle or time-sharing clients without significant interaction by users. Research questions and challenges associated with issues and limitations with the work distribution paradigm, targeted computational domain, performance metrics, and the intended class of applications to be used in this context are analyzed and discussed. A portable web services approach to master/worker parallel discrete event simulation is proposed and evaluated with subsequent optimizations to increase the efficiency of large-scale simulation execution through distributed master service design and intrinsic overhead reduction. New techniques for addressing challenges associated with optimistic parallel discrete event simulation across metacomputing such as rollbacks and message unsending with an inherently different computation paradigm utilizing master services and time windows are proposed and examined. Results indicate that a master/worker approach utilizing loosely coupled resources is a viable means for high throughput parallel discrete event simulation by enhancing existing computational capacity or providing alternate execution capability for less time-critical codes.

Les travaux présentés dans cette thèse portent sur le placement et l’ordonnancement d’applications de flots de données. On se place dans le contexte de plates-formes composées de processeurs sujets à des pannes. Dans une première partie, on considère un type particulier d’applications de flots de données: les services filtrants. On étudie l'ordonnancement de telles applications sur des plates-formes homogènes et hétérogènes, d'abord sans tenir compte des coûts de communication, puis en les incluant dans le modèle. On considère enfin l’ordonnancement d’un tel calcul sur une chaîne de processeurs. Le comportement d’un service filtrant est comparable à celui d’un calcul effectué sur un processeur non fiable: certains résultats vont être calculés, et d’autres perdus. On étudie le modèle des pannes transitoires. On veut effectuer un calcul à la fois fiable et efficace. La complexité de différentes variantes de ce problème est démontrée. Deux heuristiques sont décrites, puis comparées expérimentalement. Si les pannes transitoires sont les pannes les plus fréquemment rencontrées sur des grilles de calculs classiques, certains types de plates-formes rencontrent d’autres types de défaillances. Les grilles de volontaires sont particulièrement instables. Sur ce type de plate-forme, on veut exécuter des calculs itératifs. Cette application est constituée soit de tâches indépendantes, soit de tâches couplées, qui doivent être calculées ensemble et au même rythme. Dans chaque cas, le problème est d’abord étudié théoriquement, puis des heuristiques sontproposées, et leur performances sont comparées. / This thesis deals with the mapping and the scheduling of workflows. In this context, we consider unreliable platforms, with processors subject to failures. In a first part, we consider a particular model of streaming applications : the filtering services. In this context, we aim at minimizing period and latency. We first neglect communication costs. In this model, we study scheduling problems on homogeneous and heterogeneous platforms. Then, the impact of communication costs on scheduling problems of a filtering application is studied. Finally, we consider the scheduling problem of such an application on a chain of processors. The theoretical complexity of any variant of this problem is proved. This filtering property can model the reliability of processors. The results of some computations are successfully computed, and some other ones are lost. We consider the more frequent failure types : transient failures. We aim efficient and reliable schedules. The complexity of many variants of this problem is proved. Two heuristics are proposed and compared using using simulations. Even if transient failures are the most common failures in classical grids, some particular type of platform are more concerned by other type of problems. Desktop grids are especially unstable. In this context, we want to execute iterative applications. All tasks are executed, then a synchronization occurs, and so on. Two variants of this problem are considered : applicationsof independent tasks, and applications where all tasks need to be executed at same speed. In both cases, the problem is first theoretically studied, then heuristics are proposed and compared using simulations.