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.

Este trabalho de mestrado apresenta um estudo de Grades Computacionais e Simulações Distribuídas sobre a técnica MRIP. A partir deste estudo foi possível propor e implementar o protótipo de uma ferramenta para Gerenciamento de Experimento em Ambiente de Grade, denominada Grid Experiments Manager - GEM, organizada de forma modular podendo ser usada como um programa ou integrada com outro software, podendo ser expansível para vários middlewares de Grades Computacionais. Com a implementação também foi possível avaliar o desempenho de simulações sequenciais com aquelas executadas em cluster e em uma Grade Computacional de teste, sendo construído um benchmark que possibilitou repetir a mesma carga de trabalho para os sistemas sobre avaliação. Com os testes foi possível verificar um ganho alto no tempo de execução, quando comparadas as execuções sequenciais e em cluster, obteve-se eficiência em torno de 197% para simulações com tempo de execução baixo e 239% para aquelas com tempo de execução maior; na comparação das execuções em cluster e em grade, obteve-se os valores para eficiência de 98% e 105%, para simulações pequenas e grandes, respectivamente / This master\'s thesis presents a study of Grid Computing and Distributed Simulations using the MRIP approach. From this study was possible to design and implement the prototype of a tool for Management of Experiments in Grid Environment, called Grid Experiments Manager - GEM, which is organized in a modular way and can be used as a program or be integrated with another piece of software, being expansible to varius middlewares of Computational Grids. With its implementation was also possible to evaluate the performance of sequencial simulations executed in clusters and a Computational testbed Grid, also being implemented a benchmark which allowed repeat the same workload at the systems in evaluation. A high gain turnaround of the executions was infered with those results. When compared Sequential and Cluster executions, the eficiency was about of 197% for thin time of execution and 239% for those bigger in execution; when compared Cluster and Grid executions, the eficiency was about of 98% and 105% for thin and bigger simulations, repectivelly

Les performances des outils numériques de simulation électromagnétique de structures complexes, i.e., échelles multiples, sont souvent limitées par les ressources informatiques disponibles. De nombreux méso-centres, fermes et grilles de calcul, se créent actuellement sur les campus universitaires. Utilisant ces ressources informatiques mutualisées, ce travail de thèse s'attache à évaluer les potentialités du concept de grille de calcul (Grid Computing) pour la simulation électromagnétique de structures multi-échelles. Les outils numériques de simulation électromagnétique n'étant pas conçus pour être utilisés dans un environnement distribué, la première étape consistait donc à les modifier afin de les déployer sur une grille de calcul. Une analyse approfondie a ensuite été menée pour évaluer les performances des outils de simulation ainsi déployés sur l'infrastructure informatique. Des nouvelles approches pour le calcul électromagnétique distribué avec ces outils sont présentées et validées. En particulier, ces approches permettent la réalisation de simulation électromagnétique de structures à échelles multiples en un temps record et avec une souplesse d'utilisation. / The numerical electromagnetic tools for complex structures simulation, i.e. multi-scale, are often limited by available computation resources. Nowadays, Grid computing has emerged as an important new field, based on shared distributed computing resources of Universities and laboratories. Using these shared resources, this study is focusing on grid computing potential for electromagnetic simulation of multi-scale structure. Since the numerical simulations tools codes are not initially written for distributed environment, the first step consists to adapt and deploy them in Grid computing environment. A performance study is then realized in order to evaluate the efficiency of execution on the test-bed infrastructure. New approaches for distributing the electromagnetic computations on the grid are presented and validated. These approaches allow a very remarkable simulation time reduction for multi-scale structures and friendly-user interfaces

No description available.

No description available.

Um dos maiores desafios envolvidos no projeto de grades computacionais é o escalonamento de processos, o qual consiste no mapeamento de processos sobre os computadores disponíveis, a fim de reduzir o tempo de execução de aplicações ou maximizar a utilização de recursos. A literatura na área de Sistemas Distribuídos trata, geralmente, esses dois objetivos separadamente, dando origem às abordagens de escalonamento orientado a aplicações e orientado a recursos, respectivamente. Mais recentemente, uma nova abordagem, denominada escalonamento orientado a sistemas, tem recebido destaque, buscando otimizar ambos objetivos simultaneamente. Seguindo essas abordagens, algoritmos heurísticos e de aproximação têm sido propostos. Os heurísticos buscam por soluções de maneira eficiente sem, contudo, apresentar garantias quanto à qualidade das soluções obtidas. Em contrapartida, os algoritmos de aproximação provêm tais garantias, contudo são mais difíceis de serem projetados, o que justifica o fato de haver apenas versões simplificadas desses algoritmos para cenários de escalonamento de processos. A falta de algoritmos de aproximação adequados para abordar o problema de escalonamento de processos e a necessidade de soluções que atendam o escalonamento orientado a sistemas motivaram esta tese de doutorado que apresenta a proposta do Min Heap-based Scheduling Algorithm (MHSA), um algoritmo de aproximação para o problema de escalonamento de processos orientado a sistemas. Esse algoritmo foi baseado em um modelo de otimização matemática proposto no contexto desta tese. Esse modelo considera os comportamentos de processos e recursos a fim de quantificar a qualidade de soluções de escalonamento. O funcionamento do MHSA envolve a construção de uma árvore min-heap, em que os nós representam computadores e as chaves de ordenação correspondem aos tempos de fila, i.e., ocupação dos computadores. Apesar de esse algoritmo primordialmente reduzir o tempo de execução (ou makespan) de aplicações, essa estrutura em árvore permite que qualquer computador que ocupe o nó raiz receba cargas, o que favorece a ocupação de recursos e, portanto, sua orientação a sistemas. Esse algoritmo tem complexidade assintótica de pior caso igual a O(\'log IND. 2 m\'), em que m corresponde ao número de computadores do sistema. Sua razão de aproximação foi estudada para ambientes distribuídos heterogêneos com e sem a presença de comunicação entre processos, o que permite conhecer, a priori, o nível mínimo de qualidade alcançado por suas soluções. Experimentos foram conduzidos para avaliar o algoritmo proposto e compará-lo a outras propostas. Os resultados confirmam que o MHSA reduz o tempo dispendido na obtenção de boas soluções de escalonamento / One of the most important challenges involved in the design of grid computing systems is process scheduling, which maps applications into the available computers in attempt to reduce the application execution time, or maximize resource utilization. The literature of Distributed Systems usually deals with these two objectives separately, supporting the application-centric and the resourcecentric scheduling, respectively. More recently, a third approach referred to as system-centric scheduling has emerged which attempts to optimize both objectives in conjunction. Heuristic-based and approximation-based algorithms have been proposed to address this third type of scheduling. Heuristics aim to find good solutions at acceptable time constraints, without guaranteeing solution quality. On the other hand, approximation-based algorithms provide optimal solution bounds, however they are more difficult to design what makes them available only to simple scenarios. The need for approximation-based algorithms to support system-centric scheduling has motivated this thesis which presents Min Heap-based Scheduling Algorithm (MHSA). This approximation algorithm is based on a mathematical optimization model, also proposed in this work, which considers process and resource behaviors to measure the quality of scheduling solutions. MHSA builds a min-heap data structure in which tree nodes represent computers and sorting keys correspond to queuing times, i.e., computer workloads. Besides this algorithm primarily reduces application execution times (also referred to as makespan), its data structure allows any computer assume the root node and, consequently, receive workloads, what favors resource utilization. This algorithm has the worst-case time complexity equals to O(\'log IND. 2 m\'), in which m represents the number of system computers. Its approximation ratio was analyzed to heterogeneous distributed systems considering bag-of-tasks and communication-intensive applications. Having this ratio, we know the minimum quality level provided by every scheduling solution. Experiments were performed to compare MHSA to others. Results confirm MHSA reduces the time spent to obtain good quality scheduling solutions

This thesis focuses on selected areas of biomedical research in order to benefit from current computational infrastructures established in scientific community in european and global area. The theory of computation, parallelism and distributed computing, with focus on grid computing and cloud computing, is briefly introduced. Exchange of medical images was studied and a seamless integration of grid-based PACS system was established with the current distributed system in order to share DICOM medical images. Voice science was studied and access to real-time voice analysis application via remote desktop technology was introduced using customized protocol to transfer sound recording. This brings a possibility to access current legacy application remotely by voice specialists. The systems biology approach within domain of human physiology and pathophysiology was studied. Modeling methodology of human physiology was improved in order to build complex models based on acausal and object-oriented modeling techniques. Methods for conducting a parameter study (especially parameter estimation and parameter sweep) were introduced using grid computing and cloud computing technology. The identification of parameters gain substantial speedup by utilizing cloud computing deployment when performed on medium complex models of...

This thesis focuses on selected areas of biomedical research in order to benefit from current computational infrastructures established in scientific community in european and global area. The theory of computation, parallelism and distributed computing, with focus on grid computing and cloud computing, is briefly introduced. Exchange of medical images was studied and a seamless integration of grid-based PACS system was established with the current distributed system in order to share DICOM medical images. Voice science was studied and access to real-time voice analysis application via remote desktop technology was introduced using customized protocol to transfer sound recording. This brings a possibility to access current legacy application remotely by voice specialists. The systems biology approach within domain of human physiology and pathophysiology was studied. Modeling methodology of human physiology was improved in order to build complex models based on acausal and object-oriented modeling techniques. Methods for conducting a parameter study (especially parameter estimation and parameter sweep) were introduced using grid computing and cloud computing technology. The identification of parameters gain substantial speedup by utilizing cloud computing deployment when performed on medium complex models of...

[pt] Neste trabalho propomos usar o ALua, um mecanismo de comunicação orientado a eventos, baseado na linguagem interpretada Lua, para coordenação e desenvolvimento de aplicações paralelas distribuídas. ALua é um modelo de programação dual para aplicações paralelas distribuídas, que age como elemento de ligação, permitindo que partes pré- compiladas do programa sejam executadas em diferentes máquinas. Novas tecnologias em programação paralela, como computação em grade, e o interesse atual em computação distribuída para redes geográficas demandam novos níveis de flexibilidade, como o uso de estratégias de adaptação e a habilidade para um usuário interferir em uma computação sem a necessidade de interrompê-la. Além disso, devido a sua natureza assíncrona, a programação dirigida a eventos oferece um modelo apropriado para ambiente sujeitos a falhas e retardos, que são freqüentes no contexto de redes geográficas. Neste trabalho, mostramos que o ALua pode trazer a flexibilidade desejada, através de mecanismos de adaptação e monitoramento não só de aplicações, mas do próprio ambiente de execução, e ainda tirar proveito de sua natureza interpretada para permitir a intervenção do usuário na aplicação mesmo durante a sua execução. / [en] In this work we propose the use of Alua, an event-driven communication mechanism for coordinating and developing distributed parallel applications, based on the interpreted language Lua. Alua adopts a multilinguage programming model for distributed parallel applications, acting as a gluing element among precompiled program parts running on different machines. New developments in parallel programming, such as Grid computing, and current interest in wide-area distributed computing demand new levels of flexibility, such as the use of adaptive strategies and the ability for an user to interfer with a computation without having to stop it. Furthermore, because of its asynchronous nature, event-driven programming provides a suitable model for environments subject to failures and delays that are frequent in the context of geographically distributed computing. In this work we show that ALua can achieve the required flexibility through mechanisms for monitoring and adapting not only applications, but also the execution environment, and also exploit its interpretive nature to allow the programmer to modify the behavior of the application during its execution.

A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities. In the presence of grid users who are autonomous, rational, and intelligent, there is an overall degradation of the total efficiency of the computational grid in comparison to what can be achieved when the participating users are centrally coordinated . This loss in efficiency might arise due to an unwillingness on the part of some of the grid resource providers to either not perform completely or not perform to the fullest capability, the computational jobs of other users in the grid. In this thesis, our attention is focused on designing grid resource procurement mechanisms which a grid user can use for procuring resources in a computational grid based on bids submitted by autonomous, rational, and intelligent resource providers. Specifically, we follow a game theoretic and mechanism design approach to design three elegant, different incentive compatible procurement mechanisms for this purpose: G-DSIC (Grid-Dominant Strategy Incentive Compatible) mechanism which guarantees that truthful bidding is a best response for each resource provider, irrespective of what the other resource providers bid G-BIC (Grid-Bayesian Nash Incentive Compatible) mechanism which only guarantees that truthful bidding is a best response for each resource provider whenever all other resource providers also bid truthfully G-OPT (Grid-Optimal) mechanism which minimizes the cost to the grid user, satisfying at the same time, (1) Bayesian Incentive Compatibility (which guarantees that truthful bidding is a best response for each resource provider whenever all other resource providers also bid truthfully) and (2) Individual Rationality (which guarantees that the resource providers have non-negative payoffs if they participate in the bidding process). We evaluate the relative merits and demerits of the above three mechanisms using game theoretical analysis and numerical experiments. The mechanisms developed in this thesis are in the context of parameter sweep type of jobs, which consist of multiple homogeneous and independent tasks. We believe the use of the mechanisms proposed transcends beyond parameter sweep type of jobs and in general, the proposed mechanisms could be extended to provide a robust way of procuring resources in a computational grid where the resource providers exhibit rational and strategic behavior.