An analytical framework is developed for distributed management of large networks where each node makes locally its decisions. Two issues remain open. One is whether a distributed algorithm would result in a near-optimal management. The other is the complexity, i.e., whether a distributed algorithm would scale gracefully with a network size. We study these issues through modeling, approximation, and randomized distributed algorithms. For near-optimality issue, we first derive a global probabilistic model of network management variables which characterizes the complex spatial dependence of the variables. The spatial dependence results from externally imposed management constraints and internal properties of communication environments. We then apply probabilistic graphical models in machine learning to show when and whether the global model can be approximated by a local model. This study results in a sufficient condition for distributed management to be nearly optimal. We then show how to obtain a near-optimal configuration through decentralized adaptation of local configurations. We next derive a near-optimal distributed inference algorithm based on the derived local model. We characterize the trade-off between near-optimality and complexity of distributed and statistical management. We validate our formulation and theory through simulations.

Made available in DSpace on 2015-03-05T13:56:59Z (GMT). No. of bitstreams: 0 Previous issue date: 16 / Hewlett-Packard Brasil Ltda / Os serviços e as aplicações de computação móvel possuem exigências para o seu bom funcionamento, cujo atendimento é dificultado em virtude do ambiente altamente dinâmico em que estão inseridos. Nesse ambiente, o gerenciamento passa a ser peça chave para propiciar o funcionamento harmonioso tanto da infra-estrutura física, quanto das aplicações. As soluções existentes para essa finalidade se restringem à monitoração dos recursos de hardware, não oferecendo mecanismos para o gerenciamento das aplicações executadas nos dispositivos portáteis. Esta dissertação propõe uma arquitetura para gerenciamento de ambientes de computação móvel que permite obter e disseminar, de forma flexível, informações sobre o funcionamento dos dispositivos e, sobretudo, das aplicações. A arquitetura é baseada no padrão Web Services Distributed Management, que propõe o emprego de serviços Web na modelagem de interfaces de gerenciamento. A dissertação apresenta, ainda,resultados obtidos com a instanciação da arquitetura em um ambiente / Mobile computing services and applications pose special requirements to run properly, whose satisfaction is hampered due to the highly dynamic environment where they are executed. In this environment, management becomes an important element to enable harmonious functioning of both the mobile physical infrastructure and the applications. Current solutions are restricted to resource monitoring, and do not provide mechanisms to manage the applications running on top of the mobile devices. This work proposes an architecture to manage mobile computing environments, which allows one to obtain and disseminate, in a flexible manner, information about the devices and the applications. The architecture is based on the Web Services Distributed Management standard. This work also presents results obtained with the instantiation of the architecture in a real setup

Atualmente há grande investimento por parte das empresas em sistemas que não podem parar e, conseqüentemente, em sistemas de gerenciamento para perceber a ocorrência de falhas ou prevê-las e evitar que elas aconteçam. Para amenizar os efeitos de falhas no fornecimento de energia elétrica, que são ainda comuns no país e em boa parte do mundo, são utilizados os no-breaks. Esses equipamentos são projetados para garantir o fornecimento de energia por um tempo definido e limitado nos momentos em que a concessionária de energia elétrica interrompe esse fornecimento. O objetivo principal do presente trabalho é investigar uma solução de gerenciamento de no-breaks que possui como grande diferencial a utilização do paradigma de gerenciamento distribuído por delegação, com o uso do protocolo SNMP, apresentando as vantagens desse paradigma. Para a distribuição de tarefas, a solução utiliza a MIB Script e uma implementação deste padrão (JASMIN – Java Script MIB Implementation) de forma que toda a interatividade com a ferramenta e a automatização dos processos que compõem a solução foi implementada com o uso das ferramentas livres Apache e PHP. O trabalho apresenta uma breve descrição sobre o funcionamento dos no-breaks e destaca a UPS-MIB, MIB padrão para gerenciamento desses equipamentos, além de objetos importantes disponibilizados em outras MIBs proprietárias utilizadas para o gerenciamento de no-breaks. Além disso, o presente estudo sugere o uso de ferramentas para que o sistema de gerência seja responsável pelas estações alimentadas pelos nobreaks, ou seja, quando identificado que está próximo o momento do no-break não conseguir mais alimentar a carga, deve-se desligar a carga com segurança para que não ocorra uma perda de dados maior nem haja danos em componentes. Nota-se que o gerenciamento utilizado neste trabalho, realizado através da ferramenta desenvolvida, facilita a detecção de falhas no sistema de fornecimento de energia, de modo a interagir satisfatoriamente com no-breaks para verificar condições da rede elétrica e tomar decisões pró-ativas com o intuito de manter uma rede de computadores em funcionamento, sem que haja percalços por problemas de falta de energia. / Nowadays most companies are investing in non-stop systems, which leads to investments in management systems to know when a fault occurs or to foresee and avoid them. To reduce the effects of power supply faults (which are common in Brazil and in other parts of the world), people use UPS systems. These equipments are designed to guarantee the power supply for a pre-defined limited time, just in case the power company interrupts the supply. The main goal of this work is to present the implementation of a management application for UPS systems, which has the differential of using the management-bydelegation paradigm, based on SNMP, showing the advantages of this paradigm. The application uses Script MIB and an implementation of this standard (JASMIN - Java Script MIB Implementation) for task distribution, in such a way that all the tool interactivity and process automation was done using the free tools Apache and PHP. The work also presents a brief description about the UPS functioning, and shows UPS-MIB, the standard MIB for UPS management. Additionally the work presents some important objects available in some custom MIBs that are used for UPS management too. This study suggests the use of a set of tools to make the management system responsible for the workstations plugged on the UPS, which means that when the UPS is about to be unable to maintain the load, it should be turned off safely to avoid data losses or hardware faults. One can note that the management done in this work through the developed tool makes easy the fault detection in the power supply system. This leads to a good interaction with the UPS to verify the conditions of the power network in order to take decisions to maintain the computer network running safely.

Atualmente há grande investimento por parte das empresas em sistemas que não podem parar e, conseqüentemente, em sistemas de gerenciamento para perceber a ocorrência de falhas ou prevê-las e evitar que elas aconteçam. Para amenizar os efeitos de falhas no fornecimento de energia elétrica, que são ainda comuns no país e em boa parte do mundo, são utilizados os no-breaks. Esses equipamentos são projetados para garantir o fornecimento de energia por um tempo definido e limitado nos momentos em que a concessionária de energia elétrica interrompe esse fornecimento. O objetivo principal do presente trabalho é investigar uma solução de gerenciamento de no-breaks que possui como grande diferencial a utilização do paradigma de gerenciamento distribuído por delegação, com o uso do protocolo SNMP, apresentando as vantagens desse paradigma. Para a distribuição de tarefas, a solução utiliza a MIB Script e uma implementação deste padrão (JASMIN – Java Script MIB Implementation) de forma que toda a interatividade com a ferramenta e a automatização dos processos que compõem a solução foi implementada com o uso das ferramentas livres Apache e PHP. O trabalho apresenta uma breve descrição sobre o funcionamento dos no-breaks e destaca a UPS-MIB, MIB padrão para gerenciamento desses equipamentos, além de objetos importantes disponibilizados em outras MIBs proprietárias utilizadas para o gerenciamento de no-breaks. Além disso, o presente estudo sugere o uso de ferramentas para que o sistema de gerência seja responsável pelas estações alimentadas pelos nobreaks, ou seja, quando identificado que está próximo o momento do no-break não conseguir mais alimentar a carga, deve-se desligar a carga com segurança para que não ocorra uma perda de dados maior nem haja danos em componentes. Nota-se que o gerenciamento utilizado neste trabalho, realizado através da ferramenta desenvolvida, facilita a detecção de falhas no sistema de fornecimento de energia, de modo a interagir satisfatoriamente com no-breaks para verificar condições da rede elétrica e tomar decisões pró-ativas com o intuito de manter uma rede de computadores em funcionamento, sem que haja percalços por problemas de falta de energia. / Nowadays most companies are investing in non-stop systems, which leads to investments in management systems to know when a fault occurs or to foresee and avoid them. To reduce the effects of power supply faults (which are common in Brazil and in other parts of the world), people use UPS systems. These equipments are designed to guarantee the power supply for a pre-defined limited time, just in case the power company interrupts the supply. The main goal of this work is to present the implementation of a management application for UPS systems, which has the differential of using the management-bydelegation paradigm, based on SNMP, showing the advantages of this paradigm. The application uses Script MIB and an implementation of this standard (JASMIN - Java Script MIB Implementation) for task distribution, in such a way that all the tool interactivity and process automation was done using the free tools Apache and PHP. The work also presents a brief description about the UPS functioning, and shows UPS-MIB, the standard MIB for UPS management. Additionally the work presents some important objects available in some custom MIBs that are used for UPS management too. This study suggests the use of a set of tools to make the management system responsible for the workstations plugged on the UPS, which means that when the UPS is about to be unable to maintain the load, it should be turned off safely to avoid data losses or hardware faults. One can note that the management done in this work through the developed tool makes easy the fault detection in the power supply system. This leads to a good interaction with the UPS to verify the conditions of the power network in order to take decisions to maintain the computer network running safely.

Atualmente há grande investimento por parte das empresas em sistemas que não podem parar e, conseqüentemente, em sistemas de gerenciamento para perceber a ocorrência de falhas ou prevê-las e evitar que elas aconteçam. Para amenizar os efeitos de falhas no fornecimento de energia elétrica, que são ainda comuns no país e em boa parte do mundo, são utilizados os no-breaks. Esses equipamentos são projetados para garantir o fornecimento de energia por um tempo definido e limitado nos momentos em que a concessionária de energia elétrica interrompe esse fornecimento. O objetivo principal do presente trabalho é investigar uma solução de gerenciamento de no-breaks que possui como grande diferencial a utilização do paradigma de gerenciamento distribuído por delegação, com o uso do protocolo SNMP, apresentando as vantagens desse paradigma. Para a distribuição de tarefas, a solução utiliza a MIB Script e uma implementação deste padrão (JASMIN – Java Script MIB Implementation) de forma que toda a interatividade com a ferramenta e a automatização dos processos que compõem a solução foi implementada com o uso das ferramentas livres Apache e PHP. O trabalho apresenta uma breve descrição sobre o funcionamento dos no-breaks e destaca a UPS-MIB, MIB padrão para gerenciamento desses equipamentos, além de objetos importantes disponibilizados em outras MIBs proprietárias utilizadas para o gerenciamento de no-breaks. Além disso, o presente estudo sugere o uso de ferramentas para que o sistema de gerência seja responsável pelas estações alimentadas pelos nobreaks, ou seja, quando identificado que está próximo o momento do no-break não conseguir mais alimentar a carga, deve-se desligar a carga com segurança para que não ocorra uma perda de dados maior nem haja danos em componentes. Nota-se que o gerenciamento utilizado neste trabalho, realizado através da ferramenta desenvolvida, facilita a detecção de falhas no sistema de fornecimento de energia, de modo a interagir satisfatoriamente com no-breaks para verificar condições da rede elétrica e tomar decisões pró-ativas com o intuito de manter uma rede de computadores em funcionamento, sem que haja percalços por problemas de falta de energia. / Nowadays most companies are investing in non-stop systems, which leads to investments in management systems to know when a fault occurs or to foresee and avoid them. To reduce the effects of power supply faults (which are common in Brazil and in other parts of the world), people use UPS systems. These equipments are designed to guarantee the power supply for a pre-defined limited time, just in case the power company interrupts the supply. The main goal of this work is to present the implementation of a management application for UPS systems, which has the differential of using the management-bydelegation paradigm, based on SNMP, showing the advantages of this paradigm. The application uses Script MIB and an implementation of this standard (JASMIN - Java Script MIB Implementation) for task distribution, in such a way that all the tool interactivity and process automation was done using the free tools Apache and PHP. The work also presents a brief description about the UPS functioning, and shows UPS-MIB, the standard MIB for UPS management. Additionally the work presents some important objects available in some custom MIBs that are used for UPS management too. This study suggests the use of a set of tools to make the management system responsible for the workstations plugged on the UPS, which means that when the UPS is about to be unable to maintain the load, it should be turned off safely to avoid data losses or hardware faults. One can note that the management done in this work through the developed tool makes easy the fault detection in the power supply system. This leads to a good interaction with the UPS to verify the conditions of the power network in order to take decisions to maintain the computer network running safely.

Over the last decade, Internet-based services, such as electronic-mail, music-on-demand, and social-network services, have changed the ways we communicate and access information. Usually, the key functionality of such a service is in backend components, which are located in a data center, a facility for hosting computing systems and related equipment. This thesis focuses on two fundamental problems related to the management, dimensioning, and provisioning of such backend components. The first problem centers around resource allocation for a large-scale cloud environment. Data centers have become very large; they often contain hundreds of thousands of machines and applications. In such a data center, resource allocation cannot be efficiently achieved through a traditional management system that is centralized in nature. Therefore, a more scalable solution is needed. To address this problem, we have developed and evaluated a scalable and generic protocol for resource allocation. The protocol is generic in the sense that it can be instantiated for different management objectives through objective functions. The protocol jointly allocates CPU, memory, and network resources to applications that are hosted by the cloud. We prove that the protocol converges to a solution, if an objective function satisfies a certain property. We perform a simulation study of the protocol for realistic scenarios. Simulation results suggest that the quality of the allocation is independent of the system size, up to 100,000 machines and applications, for the management objectives considered. The second problem is related to performance modeling of a distributed key-value store. The specific distributed key-value store we focus on in this thesis is the Spotify storage system. Understanding the performance of the Spotify storage system is essential for achieving a key quality of service objective, namely that the playback latency of a song is sufficiently low. To address this problem, we have developed and evaluated models for predicting the performance of a distributed key-value store for a lightly loaded system. First, we developed a model that allows us to predict the response time distribution of requests. Second, we modeled the capacity of the distributed key-value store for two different object allocation policies. We evaluate the models by comparing model predictions with measurements from two different environments: our lab testbed and a Spotify operational environment. We found that the models are accurate in the sense that the prediction error, i.e., the difference between the model predictions and the measurements from the real systems, is at most 11%. / <p>QC 20131001</p>

<p>U radu je predstavljeno istraživanje vezano za poboljšanje performansi rada velikih nadzorno-upravljačkih sistema poput DMS-a. Ovaj cilj je postignut koordinacijom izvršavanja tokova aktivnosti, što podrazumeva efikasnu raspodelu zadataka na računarske resurse. U te svrhe razvijeni su i testirani različiti algoritmi. Ovakav pristup je obezbedio veći stepen iskorišćenja računarskih resursa, što je rezultiralo boljim performansama.</p> / <p>Thе paper presents an approach how to improve performance of larger scale distributed utility management system such as DMS. This goal is accomplished by using an intelligent workflow management. Workflows are divided into the atomic tasks which are scheduled to computing resources for execution. For these purposes various scheduling algorithms are developed and thoroughly tested. This approach has provided greater utilization of computing resources which further have resulted in better performance.</p>

Aujourd’hui, la production est souvent multi-site car les entreprises se recentrent sur leurs cœurs de métiers. Dans ce contexte, la gestion des projets est une tâche difficile car il faut prendre en compte la distribution de la décision et l’hétérogénéité qui peut exister entre les différentes applications de production des partenaires. Les Systèmes Multi-Agents, et notamment le modèle SCEP (Superviseur, Client, Environnement, Producteur), apportent une solution satisfaisante au problème de la distribution de la décision, en instaurant une coopération entre des agents responsables de la gestion des projets client et des agents représentant les sites de production distants. Néanmoins, ce modèle présente des limites à cause de sa faible capacité à communiquer et à coopérer avec des modèles et des systèmes de gestion hétérogènes ainsi qu’à sa difficulté à localiser les nouveaux partenaires. Dans ce mémoire, nous proposons une architecture distribuée et interopérable SCEP-SOA intégrant les concepts du modèle SCEP et ceux du modèle SOA (Service Oriented Architecture) qui offre des mécanismes de mise en relation des partenaires et permet des communications entre des systèmes et des applications hétérogènes. Pour garantir la bonne compréhension des informations échangées entre les partenaires, l’architecture SCEP-SOA met en œuvre une stratégie d’interopérabilité sémantique basée sur l’intégration des ontologies. Cette stratégie s’articule autour d’une ontologie globale et commune utilisée pour l’échange des informations, et des mécanismes de correspondances entre cette ontologie globale et les ontologies locales des partenaires. Cette architecture est illustrée sur un cas d’étude où l’on se focalise sur l’interopérabilité entre des applications dédiées à la planification des projets de fabrication multi-sites. / Today, the production is often multi-site because companies focus on their core competencies. In this context, projects’ management is a difficult task because it must take into account the distribution of the decision and the heterogeneity which can exist between the various production applications of the partners. The Multi-agents systems, in particular the SCEP model (Supervisor, Customer, Environment, Producer), offer a satisfactory solution for the decision distribution problem, by establishing a cooperation between agents responsible of the management of the customer projects and agents representing the remote production sites. Nevertheless, this model presents limits because of its weak ability to communicate and to cooperate with heterogeneous models and management systems, as well as in its difficulty to localize new partners. In this dissertation, we propose a distributed and interoperable architecture, SCEP-SOA, which integrates concepts of the SCEP model and SOA (Service Oriented Architecture) which offers mechanisms for putting in relation various partners and allows communication between heterogeneous systems and applications. To insure the good understanding of the information exchanged between the partners, the SCEP-SOA architecture implements a strategy of semantic interoperability based on the integration of ontologies. This strategy is based on the use of a shared global ontology for information exchange, and on mechanisms of mappings between the global ontology and the partners’ local ontologies. This architecture is illustrated by a study case in which we focus on the interoperability between applications dedicated to the planning of manufacturing multi-sites projects

Im Rahmen der Dissertation wurde ein Middlewaredienst entwickelt und realisiert. Es gestattet die dynamische Verwaltung heterogener Kontextquellen. Das zugrunde liegende Komponentenmodell selbstbeschreibender Context Provieder ermöglicht die lose Kopplung von Kontextquellen und -senken. Es wird durch Filter- und Konverterkomponenten zur generischen Providersselektion anhand domänenspezifischer Merkmale ergänzt. Die Kopplung der verteilten Dienstinstanzen erfolgt durch ein hybrides Peer-to-Peer-System. Dies trägt der Heterogenität der Endgeräte Rechnung, und erlaubt die skalierbare , verteilte Verwaltung von Kontextquellen in globalen Szenarien.