Selection research often examines whether adverse impact can be reduced/eliminated from employment practices. Such research, however, largely ignores the influence of minority availability issues (i.e., the number of minorities who apply and the number of minorities who accept a job offer); three general factors comprise minority availability: the missing applicant problem, targeted recruitment and job refusal rates. As minority availability issues have not been systematically addressed in the broader literature, the purpose of this study was twofold: 1) to highlight the importance of and explicate a comprehensive description of their potential effects on adverse impact and 2) to demonstrate such effects through monte carlo simulations. Specifically, simulations were used to examine issues related to the level effects and covariance effects of minority availability on adverse impact. Therefore, an iterative process was used whereby minority availability factors were manipulated to produce combinations that meaningfully affect adverse impact; the goal was to conduct as many simulations as necessary to establish a reliable pattern of the effects of minority availability on adverse impact. Simulation results suggest minority availability issues can influence the detection of adverse impact. In fact, minority availability issues may hinder efforts to reduce adverse impact in some selection contexts. Implications of these results are discussed. / Ph. D.

Atualmente, os computadores pessoais são equipados com processadores de vários núcleos possuem alto poder de processamento. Algumas pesquisas realizadas indicam que estes mesmos computadores permanecem em média 80% do tempo ociosos. A união destas duas informações fomenta à pesquisa de um tipo de computação distribuída, conhecida como grades oportunistas. A utilização de um ambiente de grade oportunista permite a implementação de um computador paralelo capaz de processar grandes volumes de tarefas, com baixo custo financeiro. Apesar do baixo custo financeiro proporcionado por este ambiente, tem-se um alto custo computacional devido a utilização de recursos não dedicados. Estes ambientes sofrem de um problema que tem servido de objeto de diversas pesquisas, a indisponibilidade de recursos. O problema da indisponibilidade de recursos deve-se a volatilidade deste recursos. A volatilidade deve-se a liberdade de tornarem-se ociosos ou ocupados a qualquer instante, sem a necessidade de aviso prévio ao servidor. A volatilidade é a responsável por gerar uma série de interrupções de execução de tarefas e a consequente necessidade de novos escalonamentos desta mesma tarefa. Para reduzir o impacto deste problema, o escalonador de tarefas poderia considerar informações sobre o comportamento futuro dos recursos, ou seja, poderia considerar informações provenientes de um preditor de disponibilidade. Este trabalho propõe o modelo AvSchedP, modelo este capaz de realizar a previsão de disponibilidade baseando-se em dados binários e de forma dinâmica. O modelo AvSchedP foi projetado para ser integrado a escalonadores de grades oportunistas. O AvSchedP foi implementado e diversos experimentos foram realizados. Os experimentos foram realizados para avaliar a qualidade de previsão e a adequação do modelo a um sistema de grade oportunista, o XtremWeb. Os resultados obtidos demonstram que o AvSchedP apresenta bons resultados ao realizar previsões de duração de disponibilidade em ambientes formados por recursos voláteis. / Nowadays, personal computers have multi-core processors with high processing power. Some researches show that these computers remain 80 % of the time idle. The union of these two information instigates a kind of distributed computing, known as opportunistic grids. The opportunistic grid environment usage allows the implementation of a parallel computer like, capable of processing large volumes of tasks, with a low cost. Despite the low financial cost, it has a high computational cost due to use of nondedicated resources. Opportunistic grid environments have a problem that has been the subject of several researches: the resource availability. The resource availability problem is caused by the volatility of resources. The resource is free to become idle or busy at any time without notice the server. The volatility is responsible for generating a lot of task execution interruptions and consequently, task rescheduling. To alleviate this problem, the task scheduler could consider information about the future behavior of resources provided by an avalibility predictor. This paper proposes a new model, called AvSchedP , to perform dynamic resource availability prediction based on binary data. The model was designed to be integrated into opportunistic grid schedulers. The AvSchedP was implemented and several experiments were performed. These experiments were conducted to evaluate the prediction quality and the model integrability to the XtremWeb opportunistic grid system. The results show that the AvSchedP has good results when making availability duration predictions for environments composed by volatile resources.

This thesis describes development and verification of application for support of HACMP clustering education in IBM AIX OS. In fact, development of this application can help students to understand basic functions of this type of cluster. Users, eventually, can verify full function of input configuration cluster in real environment. Ultimately, it can simulate basic cluster function and create basic configuration file, which can be direct apply to operational installation of HACMP software of version 5.3.

Atualmente, os computadores pessoais são equipados com processadores de vários núcleos possuem alto poder de processamento. Algumas pesquisas realizadas indicam que estes mesmos computadores permanecem em média 80% do tempo ociosos. A união destas duas informações fomenta à pesquisa de um tipo de computação distribuída, conhecida como grades oportunistas. A utilização de um ambiente de grade oportunista permite a implementação de um computador paralelo capaz de processar grandes volumes de tarefas, com baixo custo financeiro. Apesar do baixo custo financeiro proporcionado por este ambiente, tem-se um alto custo computacional devido a utilização de recursos não dedicados. Estes ambientes sofrem de um problema que tem servido de objeto de diversas pesquisas, a indisponibilidade de recursos. O problema da indisponibilidade de recursos deve-se a volatilidade deste recursos. A volatilidade deve-se a liberdade de tornarem-se ociosos ou ocupados a qualquer instante, sem a necessidade de aviso prévio ao servidor. A volatilidade é a responsável por gerar uma série de interrupções de execução de tarefas e a consequente necessidade de novos escalonamentos desta mesma tarefa. Para reduzir o impacto deste problema, o escalonador de tarefas poderia considerar informações sobre o comportamento futuro dos recursos, ou seja, poderia considerar informações provenientes de um preditor de disponibilidade. Este trabalho propõe o modelo AvSchedP, modelo este capaz de realizar a previsão de disponibilidade baseando-se em dados binários e de forma dinâmica. O modelo AvSchedP foi projetado para ser integrado a escalonadores de grades oportunistas. O AvSchedP foi implementado e diversos experimentos foram realizados. Os experimentos foram realizados para avaliar a qualidade de previsão e a adequação do modelo a um sistema de grade oportunista, o XtremWeb. Os resultados obtidos demonstram que o AvSchedP apresenta bons resultados ao realizar previsões de duração de disponibilidade em ambientes formados por recursos voláteis. / Nowadays, personal computers have multi-core processors with high processing power. Some researches show that these computers remain 80 % of the time idle. The union of these two information instigates a kind of distributed computing, known as opportunistic grids. The opportunistic grid environment usage allows the implementation of a parallel computer like, capable of processing large volumes of tasks, with a low cost. Despite the low financial cost, it has a high computational cost due to use of nondedicated resources. Opportunistic grid environments have a problem that has been the subject of several researches: the resource availability. The resource availability problem is caused by the volatility of resources. The resource is free to become idle or busy at any time without notice the server. The volatility is responsible for generating a lot of task execution interruptions and consequently, task rescheduling. To alleviate this problem, the task scheduler could consider information about the future behavior of resources provided by an avalibility predictor. This paper proposes a new model, called AvSchedP , to perform dynamic resource availability prediction based on binary data. The model was designed to be integrated into opportunistic grid schedulers. The AvSchedP was implemented and several experiments were performed. These experiments were conducted to evaluate the prediction quality and the model integrability to the XtremWeb opportunistic grid system. The results show that the AvSchedP has good results when making availability duration predictions for environments composed by volatile resources.

Atualmente, os computadores pessoais são equipados com processadores de vários núcleos possuem alto poder de processamento. Algumas pesquisas realizadas indicam que estes mesmos computadores permanecem em média 80% do tempo ociosos. A união destas duas informações fomenta à pesquisa de um tipo de computação distribuída, conhecida como grades oportunistas. A utilização de um ambiente de grade oportunista permite a implementação de um computador paralelo capaz de processar grandes volumes de tarefas, com baixo custo financeiro. Apesar do baixo custo financeiro proporcionado por este ambiente, tem-se um alto custo computacional devido a utilização de recursos não dedicados. Estes ambientes sofrem de um problema que tem servido de objeto de diversas pesquisas, a indisponibilidade de recursos. O problema da indisponibilidade de recursos deve-se a volatilidade deste recursos. A volatilidade deve-se a liberdade de tornarem-se ociosos ou ocupados a qualquer instante, sem a necessidade de aviso prévio ao servidor. A volatilidade é a responsável por gerar uma série de interrupções de execução de tarefas e a consequente necessidade de novos escalonamentos desta mesma tarefa. Para reduzir o impacto deste problema, o escalonador de tarefas poderia considerar informações sobre o comportamento futuro dos recursos, ou seja, poderia considerar informações provenientes de um preditor de disponibilidade. Este trabalho propõe o modelo AvSchedP, modelo este capaz de realizar a previsão de disponibilidade baseando-se em dados binários e de forma dinâmica. O modelo AvSchedP foi projetado para ser integrado a escalonadores de grades oportunistas. O AvSchedP foi implementado e diversos experimentos foram realizados. Os experimentos foram realizados para avaliar a qualidade de previsão e a adequação do modelo a um sistema de grade oportunista, o XtremWeb. Os resultados obtidos demonstram que o AvSchedP apresenta bons resultados ao realizar previsões de duração de disponibilidade em ambientes formados por recursos voláteis. / Nowadays, personal computers have multi-core processors with high processing power. Some researches show that these computers remain 80 % of the time idle. The union of these two information instigates a kind of distributed computing, known as opportunistic grids. The opportunistic grid environment usage allows the implementation of a parallel computer like, capable of processing large volumes of tasks, with a low cost. Despite the low financial cost, it has a high computational cost due to use of nondedicated resources. Opportunistic grid environments have a problem that has been the subject of several researches: the resource availability. The resource availability problem is caused by the volatility of resources. The resource is free to become idle or busy at any time without notice the server. The volatility is responsible for generating a lot of task execution interruptions and consequently, task rescheduling. To alleviate this problem, the task scheduler could consider information about the future behavior of resources provided by an avalibility predictor. This paper proposes a new model, called AvSchedP , to perform dynamic resource availability prediction based on binary data. The model was designed to be integrated into opportunistic grid schedulers. The AvSchedP was implemented and several experiments were performed. These experiments were conducted to evaluate the prediction quality and the model integrability to the XtremWeb opportunistic grid system. The results show that the AvSchedP has good results when making availability duration predictions for environments composed by volatile resources.

Olika tekniker för att beräkna webbtjänsters upp-tid har länge vart en viktig huvuddel till att förbättra samverkan och systemintegration. Tillgänglighet är den mest vanliga metoden när man presenterar dessa uträkningar där målet är främst att locka kunder, men kanske ännu viktigare att med hjälp av resultatet se hur man som utvecklare eller användare kan förbättra webbtjänsten. Tillgänglighet varierar mycket beroende på vad det är för typ av tjänst, företag eller användning, och ett vanligt problem är definitionen av vad tillgänglighet egentligen är. Det här arbetet kommer att ge dess läsare en introduktion till området tillgänglighet, och förklara varför den varierar, vilket är grunden till teorin bakom interaktionsdesign. Tillgänglighet är ett mätvärde som kan beräknas på många olika sätt, och generar ett resultat som är komplext för användarna som tar del av det. Målet blir främst att ge läsaren en uppfattning och riktlinjer om hur man kan definiera den upplevda tillgängligheten baserat på ett systems tillgänglighet, men också presentera metoder till att definiera, beräkna och presentera variabeln i en användarvänlig procedur. Resultatet kommer i sin tur att bestå av en kostnadseffektiv modell för upplevd tillgänglighet och bli testad på ett multinationellt företag. / Techniques of measuring web-service uptime have always been a key metric to improve interaction and system integration. Availability is a usual metric in the field of statistics where the goal is to attract customers, but perhaps more importantly the users and providers which can improve the services through this metric. Since availability differs a lot depending on type of service, company and usage, a common problem is to define what availability really is. The thesis will give the readers an introduction to availability, and also explain the reasons why it may vary, which is the theory of interaction design behind a service. Though availability is a metric that can be calculated through many different ways, the result is very complex to understand for the public that is interested in it. The goal of this thesis is to give the readers an understanding and guidelines of how to define perceived availability based on the system availability, but also present a method of defining, calculating and present the metric in a user-friendly procedure. The result will in turn consist of a cost-effective model for perceived availability and be tested at a multinational company.

京都大学 / 0048 / 新制・課程博士 / 博士(地球環境学) / 甲第22905号 / 地環博第206号 / 新制||地環||40(附属図書館) / 京都大学大学院地球環境学舎地球環境学専攻 / (主査)教授 舟川 晋也, 准教授 渡邉 哲弘, 准教授 舘野 隆之輔 / 学位規則第4条第1項該当 / Doctor of Global Environmental Studies / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(農学) / 甲第21968号 / 農博第2358号 / 新制||農||1070(附属図書館) / 学位論文||R1||N5219(農学部図書室) / 京都大学大学院農学研究科地域環境科学専攻 / (主査)教授 北山 兼弘, 教授 小杉 緑子, 教授 舟川 晋也 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DGAM

Cellular networks in Africa have grown exponentially over the past 10 years and their data centres (DCs) on average consume 3 MW of electrical power. They require a reliable electrical power supply and can have a downtime loss of over a million dollars per hour. Power quality, reliability and availability have emerged as key issues for the successful operation of a data centre. Investigations are carried out into emerging technologies and their application in data centre power distribution systems for cellular networks in Africa. Best practices are applied to develop a power distribution system (PDS) with the objective of achieving optimal reliability and availability. Analytical techniques are applied to determine and compare the reliability and availability of various power systems. Minimal cut set simulations identify system weak points and confirm component selection. Components’ inherent characteristics (CIC) and system connectivity topology (SCT) are key factors in the improvement of data centre availability. The analysis practices can be used by engineers and managers as a basis for informed decision making in determining power system reliability and the availability of an existing or a new data centre design. Weak points in the PDS of a data centre causing downtime are identified through analysis, and accurate solutions can be determined to prevent or minimise downtime. System connectivity topology (SCT) techniques were identified that could increase the reliability and availability of data centres for cellular networks in Africa. These techniques include multiple incomers from the utility company, redundancy levels of critical equipment and parallel distribution paths. Two case studies were carried out on data centres for a cellular network, one in Nigeria and one in Cameroon. The reliability and availability of both data centres was improved, with substantial reduction in downtime per year. The outcome of the case studies shows the importance of designing and implementing the power distribution system with sufficient levels of redundancy for critical equipment, and parallel distribution paths. / MSc (Engineering Sciences in Nuclear Engineering), North-West University, Potchefstroom Campus, 2014

Cellular networks in Africa have grown exponentially over the past 10 years and their data centres (DCs) on average consume 3 MW of electrical power. They require a reliable electrical power supply and can have a downtime loss of over a million dollars per hour. Power quality, reliability and availability have emerged as key issues for the successful operation of a data centre. Investigations are carried out into emerging technologies and their application in data centre power distribution systems for cellular networks in Africa. Best practices are applied to develop a power distribution system (PDS) with the objective of achieving optimal reliability and availability. Analytical techniques are applied to determine and compare the reliability and availability of various power systems. Minimal cut set simulations identify system weak points and confirm component selection. Components’ inherent characteristics (CIC) and system connectivity topology (SCT) are key factors in the improvement of data centre availability. The analysis practices can be used by engineers and managers as a basis for informed decision making in determining power system reliability and the availability of an existing or a new data centre design. Weak points in the PDS of a data centre causing downtime are identified through analysis, and accurate solutions can be determined to prevent or minimise downtime. System connectivity topology (SCT) techniques were identified that could increase the reliability and availability of data centres for cellular networks in Africa. These techniques include multiple incomers from the utility company, redundancy levels of critical equipment and parallel distribution paths. Two case studies were carried out on data centres for a cellular network, one in Nigeria and one in Cameroon. The reliability and availability of both data centres was improved, with substantial reduction in downtime per year. The outcome of the case studies shows the importance of designing and implementing the power distribution system with sufficient levels of redundancy for critical equipment, and parallel distribution paths. / MSc (Engineering Sciences in Nuclear Engineering), North-West University, Potchefstroom Campus, 2014