A demanda por alto desempenho é um desafio enfrentado por diversas aplicações científicas. Nesse sentido, ambientes para processamento distribuído, tais como, clusters e grades computacionais, têm sido desenvolvidos para prover suporte ao uso de diversos recursos simultaneamente para uma mesma aplicação. Aplicações computacionalmente intensivas são organizadas em workflows e executadas com suporte de middlewares para abstrair a complexidade de uso de tais ambientes. Em ambientes de grade computacional, a execução de workflows contendo tarefas sequenciais e tarefas com paralelismo interno, obtendo bom desempenho, é um desafio, devido à heterogeneidade e comportamento dinâmico do ambiente. Nesse sentido, o escalonamento de workflows em ambientes de grade computacional é essencial. O problema de escalonamento de tarefas, em sua forma geral, é NPCompleto, dessa forma, o estudo do escalonamento de workflows em ambientes de grade computacional é fundamental para aprimorar a execução de aplicações computacionalmente intensivas. O objetivo dessa tese é propor estratégias de escalonamento de workflows, que exploram os seguintes aspectos: Avaliação da possibilidade de executar cada tarefa com paralelismo interno usando recursos de múltiplos clusters; Adaptação de planos de escalonamento no momento da submissão de novos workflows. Foram desenvolvidas duas estratégias: a primeira é uma estratégia para escalonamento estático de workflows, que considera o ambiente dedicado a execução de um workflow. A segunda foi desenvolvida para ser utilizada em conjunto com a primeira, para melhorar o tempo de resposta de múltiplos workflows que podem ser submetidos em diferentes momentos. As estratégias propostas foram avaliadas em um ambiente de simulação. / The demand for high performance is a common problem in many scientific applications. In this sense, distributed processing environments such as cluster, grid computing and multi-cluster environments have been developed to provide support for the use of several resources simultaneously for the same application. Computationally intensive applications are structured as workflows and executed with the support of middleware to abstract the complexity of using such environments. In grid computing environments the execution of workflows containing sequential and parallel tasks, with good performance is a challenge due to the heterogeneity and dynamic behavior of the environment. In this sense, the scheduling of workflows on grid computing environments is essential. The task scheduling problem in its general form is NP-Complete, in this sense, the study concerning workflow scheduling in grid computing environments is fundamental to improve the performance of computationally intensive applications. The aim of this thesis is to propose strategies for scheduling workflows that exploit the following aspects: Explore the possibility of performing single parallel tasks using multiple clusters; Adaptation plans escalation in accordance with the submission of new workflows. Two strategies were developed: the first one is a strategy for static scheduling of workflows, which considers a dedicated environment to the execution of a workflow. The second one was developed to use in conjunction with the first one, in order to improve the response time of multiple workflows that can be submitted at different times. The proposed strategies were evaluated in a simulation environment.

A demanda por alto desempenho é um desafio enfrentado por diversas aplicações científicas. Nesse sentido, ambientes para processamento distribuído, tais como, clusters e grades computacionais, têm sido desenvolvidos para prover suporte ao uso de diversos recursos simultaneamente para uma mesma aplicação. Aplicações computacionalmente intensivas são organizadas em workflows e executadas com suporte de middlewares para abstrair a complexidade de uso de tais ambientes. Em ambientes de grade computacional, a execução de workflows contendo tarefas sequenciais e tarefas com paralelismo interno, obtendo bom desempenho, é um desafio, devido à heterogeneidade e comportamento dinâmico do ambiente. Nesse sentido, o escalonamento de workflows em ambientes de grade computacional é essencial. O problema de escalonamento de tarefas, em sua forma geral, é NPCompleto, dessa forma, o estudo do escalonamento de workflows em ambientes de grade computacional é fundamental para aprimorar a execução de aplicações computacionalmente intensivas. O objetivo dessa tese é propor estratégias de escalonamento de workflows, que exploram os seguintes aspectos: Avaliação da possibilidade de executar cada tarefa com paralelismo interno usando recursos de múltiplos clusters; Adaptação de planos de escalonamento no momento da submissão de novos workflows. Foram desenvolvidas duas estratégias: a primeira é uma estratégia para escalonamento estático de workflows, que considera o ambiente dedicado a execução de um workflow. A segunda foi desenvolvida para ser utilizada em conjunto com a primeira, para melhorar o tempo de resposta de múltiplos workflows que podem ser submetidos em diferentes momentos. As estratégias propostas foram avaliadas em um ambiente de simulação. / The demand for high performance is a common problem in many scientific applications. In this sense, distributed processing environments such as cluster, grid computing and multi-cluster environments have been developed to provide support for the use of several resources simultaneously for the same application. Computationally intensive applications are structured as workflows and executed with the support of middleware to abstract the complexity of using such environments. In grid computing environments the execution of workflows containing sequential and parallel tasks, with good performance is a challenge due to the heterogeneity and dynamic behavior of the environment. In this sense, the scheduling of workflows on grid computing environments is essential. The task scheduling problem in its general form is NP-Complete, in this sense, the study concerning workflow scheduling in grid computing environments is fundamental to improve the performance of computationally intensive applications. The aim of this thesis is to propose strategies for scheduling workflows that exploit the following aspects: Explore the possibility of performing single parallel tasks using multiple clusters; Adaptation plans escalation in accordance with the submission of new workflows. Two strategies were developed: the first one is a strategy for static scheduling of workflows, which considers a dedicated environment to the execution of a workflow. The second one was developed to use in conjunction with the first one, in order to improve the response time of multiple workflows that can be submitted at different times. The proposed strategies were evaluated in a simulation environment.

With the rise of microservice architectures, there is a need for an orchestration tool to manage containers. Kubernetes has emerged as one of the most popular alternatives, adopting widespread usage. But managing multiple Kubernetes clusters on its own have proven to be a challenging task. This difficulty has given rise to multiple cloud based alternatives which help streamline the managing process of a cluster environment and helps maintain an extreme high availability environment that is hard to replicate in an on premise environment. Using these cloud based platforms for hosting and managing ones system is great, but alleviating control of a system to a cloud provider masquerades any illicit behaviour performed on or through the system. The scope of this thesis is on examining optional designs that will automate the process of executing a geographical failover between different locations to better sustain an on premise fault tolerant kubernetes environment. There already exists multiple tools in the area of kubernetes service mesh, but their focus is not primarily on increasing system resilience but to increase security, observability and performance. Linkerd is a sidecar oriented service mesh which supports geographical failover by manually announcing individual services between cluster(s) mirror gateways. Cilium offers an Container Networking Interface (CNI) which performs routing through eBPF and allows for seamless failover between clusters by managing cross cluster service endpoints. Both of the mentioned service mesh providers handle failover from inside the kubernetes cluster. The contributions includes two new peer-to-peer designs that focus on external cluster geographical failover - both designs are compatible with preexisting kubernetes clusters without internal modifications. A fully repli-cated design was then realised into a proof of concept (POC), and tested against a Cilium multi cluster environment on the metric of north to south traffic latency. Due to the nature of the underlying hardware, the tests showed that the POC can be used for external geographical failover and it showed potential performance capabilities in a limited lab scale. As the purpose of this thesis was not to determine the traffic throughput of a geographical failover solution; but to examine different approaches automatic geographical failover can be implemented, the tests were a success. Therefore, this thesis can conclude that there exists several working solutions, and the POC have shown that there are still undiscovered and unimplemented solutions to explore.

Computer networks can be classified into two broad categories: wired networks and wireless networks, according to the hardware and software technologies used to interconnect the individual devices. Wired interconnection networks are hardware fabrics supporting communications between individual processors in highperformance computing systems (e.g., multi-computer systems and cluster systems). On the other hand, due to the rapid development of wireless technologies, wireless networks have emerged and become an indispensable part for people's lives. The integration of different wireless technologies is an effective approach to accommodate the increasing demand of the users to communicate with each other and access the Internet. This thesis aims to investigate the performance of wired interconnection networks and integrated wireless networks under the realistic working conditions. Traffic patterns have a significant impact on network performance. A number of recent measurement studies have convincingly demonstrated that the traffic generated by many real-world applications in communication networks exhibits bursty arrival nature and the message destinations are non-uniformly distributed. Analytical models for the performance evaluation of wired interconnection networks and integrated wireless networks have been widely reported. However, most of these models are developed under the simplified assumption of non-bursty Poisson process with uniformly distributed message destinations. To fill this gap, this thesis first presents an analytical model to investigate the performance of wired interconnection networks in multi-computer systems. Secondly, the analytical models for wired interconnection networks in multi-cluster systems are developed. Finally, this thesis proposes analytical models to evaluate the end-to-end delay and throughput of integrated wireless local area networks and wireless mesh networks. These models are derived when the networks are subject to bursty traffic with non-uniformly distributed message destinations which can capture the burstiness of real-world network traffic in the both temporal domain and spatial domain. Extensive simulation experiments are conducted to validate the accuracy of the analytical models. The models are then used as practical and cost-effective tools to investigate the performance of heterogeneous wired or wireless networks under the traffic patterns exhibited by real-world applications.

La complexité des architectures de communication avioniques ne cesse de croître avec l’augmentation du nombre des terminaux interconnectés et l’expansion de la quantité des données échangées. Afin de répondre aux besoins émergents en terme de bande passante, latence et modularité, l’architecture de communication avionique actuelle consiste à utiliser le réseau AFDX (Avionics Full DupleX Switched Ethernet) pour connecter les calculateurs et utiliser des bus d’entrée/sortie (par exemple le bus CAN (Controller Area Network)) pour connecter les capteurs et les actionneurs. Les réseaux ainsi formés sont connectés en utilisant des équipements d’interconnexion spécifiques, appelés RDC (Remote Data Concentrators) et standardisé sous la norme ARINC655. Les RDCs sont des passerelles de communication modulaires qui sont reparties dans l’avion afin de gérer l’hétérogénéité entre le réseau cœur AFDX et les bus d’entrée/sortie. Certes, les RDCs permettent d’améliorer la modularité du système avionique et de réduire le coût de sa maintenance; mais, ces équipements sont devenus un des défis majeurs durant la conception de l’architecture avionique afin de garantir les performances requises du système. Les implémentations existantes du RDC effectuent souvent une translation direct des trames et n’implémentent aucun mécanisme de gestion de ressources. Or, une utilisation efficace des ressources est un besoin important dans le contexte avionique afin de faciliter l’évolution du système et l’ajout de nouvelles fonctions. Ainsi, l’objectif de cette thèse est la conception et la validation d’un RDC optimisé implémentant des mécanismes de gestion des ressources afin d’améliorer les performances de l’architecture de communication avionique tout en respectant les contraintes temporelles du système. Afin d’atteindre cet objectif, un RDC pour les architectures réseaux de type CAN-AFDX est conçu, intégrant les fonctions suivantes: (i) groupement des trames appliqué aux flux montants, i.e., flux générés par les capteurs et destinés à l’AFDX, pour minimiser le coût des communication sur l’AFDX; (ii) la régulation des flux descendants, i.e., flux générés par des terminaux AFDX et destinés aux actionneurs, pour réduire les contentions sur le bus CAN. Par ailleurs, notre RDC permet de connecter plusieurs bus CAN à la fois tout en garantissant une isolation entre les flux. Par la suite, afin d’analyser l’impact de ce nouveau RDC sur les performances du système avionique, nous procédons à la modélisation de l’architecture CAN-AFDX, et particulièrement le RDC et ses nouvelles fonctions. Ensuite, nous introduisons une méthode d’analyse temporelle pour calculer des bornes maximales sur les délais de bout en bout et vérifier le respect des contraintes temps-réel. Plusieurs configurations du RDC peuvent répondre aux exigences du système avionique tout en offrant des économies de ressources. Nous procédons donc au paramétrage du RDC afin de minimiser la consommation de bande passante sur l’AFDX tout en respectant les contraintes temporelles. Ce problème d’optimisation est considéré comme NP-complet, et l’introduction des heuristiques adéquates s’est avérée nécessaire afin de trouver la meilleure configuration possible du RDC. Enfin, les performances de ce nouveau RDC sont validées à travers une architecture CAN-AFDX réaliste, avec plusieurs bus CAN et des centaines de flux échangés. Différents niveaux d’utilisation des bus CAN ont été considérés et les résultats obtenus ont montré l’efficacité de notre RDC à améliorer la gestion des ressources du système avionique tout en respectant les contraintes temporelles de communication. En particulier, notre RDC offre une réduction de la bande passante AFDX allant jusqu’à 40% en comparaison avec le RDC actuellement utilisé. / The aim of my thesis is to provide a resources-efficient gateway to connect Input/Output (I/O) CAN buses to a backbone network based on AFDX technology, in modern avionics communication architectures. Currently, the Remote Data Concentrator (RDC) is the main standard for gateways in avionics; and the existing implementations do not integrate any resource management mechanism. To handle these limitations, we design an enhanced CAN-AFDX RDC integrating new functions: (i) Frame Packing (FP) allowing to reduce communication overheads with reference to the currently used "1 to 1" frame conversion strategy; (ii) Hierarchical Traffic Shaping (HTS) to reduce contention on the CAN bus. Furthermore, our proposed RDC allows the connection of multiple I/O CAN buses to AFDX while guaranteeing isolation between different criticality levels, using a software partitioning mechanism. To analyze the performance guarantees offered by our proposed RDC, we considered two metrics: the end-to-end latency and the induced AFDX bandwidth consumption. Furthermore, an optimization process was proposed to achieve an optimal configuration of our proposed RDC, i.e., minimizing the bandwidth utilization while meeting the real-time constraints of communication. Finally, the capacity of our proposed RDC to meet the emerging avionics requirements has been validated through a realistic avionics case study.

Le système de fabrication des plaquettes de silicium (wafer) est la partie la plus complexe et la plus coûteuse du processus de fabrication des semi-conducteurs et son ordonnancement pour la production a un impact significatif sur la rentabilité économique. Le système d’outils Multi-cluster pour la fabrication de plaquettes est un système de type multi-boucles, largement utilisé dans la fabrication de plaquettes de 300 mm et 450 mm. Le problème d’ordonnancement dans ce système de production présente des caractéristiques pour les modèles de flux de plaquettes compliqué, des contraintes résidentielles strictes et des conflits de ressources à gérer, ce qui rend le problème très complexe. Dans cette thèse, l'outil multi-cluster est étudié et les recherches se concentrent principalement sur les caractéristiques des contraintes sur le temps de séjour, les contraintes sur les ressources utilisés et les flux plaquettes de silicium. Plus particulièrement, cette thèse traite trois problèmes d'ordonnancement: le problème d'ordonnancement cyclique unitaire pour un flux unique de plaquettes, le problème d'ordonnancement cyclique multi-unitaires dans un modèle de flux unique de plaquettes et le problème d'ordonnancement non-cyclique. Pour résoudre ces problèmes, des modèles robustes sont développés ainsi que certains algorithmes heuristiques efficaces sont construits pour atteindre les objectifs. L'objectif principal étant d'améliorer la performance des outils multi-cluster et d'augmenter le rendement des flux des plaquettes de silicium. Des tests de simulation et des analyses sont effectuées afin d’évaluer la performance des algorithmes proposés. Les résultats montrent la stabilité et l'efficacité de ces algorithmes. / Multi-cluster tool is a highly automated and costly wafer fabrication system with multi-loop coupling structure, and scheduling of such equipment directly affects the overall efficiency of semiconductor manufacturing enterprises. Multi-cluster tools scheduling problem has the features of large scale, complex wafer flow patterns, strict residency time constraints and intense resource conflict, which are significantly different from any other manufacturing system. Since the existing literatures have proved that most of the wafer fabrication systems scheduling problems are NP-hard, it’s difficult to obtain the optimal solution by using exact algorithms. Thus, how to develop an efficient heuristic algorithm to solve the multi-cluster tools scheduling problem attracts considerable attention both in academia and in industry. After reviewing the literatures, it is found that the research on the cyclic scheduling problem of multi-cluster tools rarely takes into account the characteristics of residency constraints. The scale of the object is limited to three single cluster tools, and the proposed scheduling methods are mostly mathematical programming and simple scheduling rules. Therefore, in this thesis, the multi-cluster tool is studied and our research mainly focuses on the characteristics of residency constraints, resource constraints and wafer flow patterns. Based on the descriptions of research domains, some solid models are developed for different scheduling problems and some efficient heuristic algorithms are constructed to realize the objectives. To deal with the problem, different approaches are proposed: A non-linear mixed-integer programming model, a two-stage = approximate-optimal scheduling algorithm, and a chaos-based particle swarm optimization-tabu search hybrid heuristic algorithm. Simulation experiments and analysis demonstrate the effectiveness of these algorithms. Results show the stability and efficiency of proposed algorithms.

Computer networks can be classified into two broad categories: wired networks and wireless networks, according to the hardware and software technologies used to interconnect the individual devices. Wired interconnection networks are hardware fabrics supporting communications between individual processors in highperformance computing systems (e.g., multi-computer systems and cluster systems). On the other hand, due to the rapid development of wireless technologies, wireless networks have emerged and become an indispensable part for people¿s lives. The integration of different wireless technologies is an effective approach to accommodate the increasing demand of the users to communicate with each other and access the Internet. This thesis aims to investigate the performance of wired interconnection networks and integrated wireless networks under the realistic working conditions. Traffic patterns have a significant impact on network performance. A number of recent measurement studies have convincingly demonstrated that the traffic generated by many real-world applications in communication networks exhibits bursty arrival nature and the message destinations are non-uniformly distributed. Analytical models for the performance evaluation of wired interconnection networks and integrated wireless networks have been widely reported. However, most of these models are developed under the simplified assumption of non-bursty Poisson process with uniformly distributed message destinations. To fill this gap, this thesis first presents an analytical model to investigate the performance of wired interconnection networks in multi-computer systems. Secondly, the analytical models for wired interconnection networks in multi-cluster systems are developed. Finally, this thesis proposes analytical models to evaluate the end-to-end delay and throughput of integrated wireless local area networks and wireless mesh networks. These models are derived when the networks are subject to bursty traffic with non-uniformly distributed message destinations which can capture the burstiness of real-world network traffic in the both temporal domain and spatial domain. Extensive simulation experiments are conducted to validate the accuracy of the analytical models. The models are then used as practical and cost-effective tools to investigate the performance of heterogeneous wired or wireless networks under the traffic patterns exhibited by real-world applications.