Orientador: Marco Aurelio Amaral Henriques / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-06T23:04:47Z (GMT). No. of bitstreams: 1 Herai_RobertoHirochi_M.pdf: 4053882 bytes, checksum: c48bdb80fdf9c0bbdb2249a62b3a2345 (MD5) Previous issue date: 2005 / Resumo: A predição de desempenho é um importante mecanismo para avaliar a utilização de recursos e estimar o tempo de execução de aplicações em sistemas paralelos. Este trabalho apresenta uma ferramenta que permite criar modelos que representam características da aplicação e dos computadores utilizados para processá-la. Tais modelos são combinados para gerar um modelo de desempenho mais abrangente, cuja análise permite obter estimativas de tempo de execução que contemplem fatores de atraso tais como operações aritméticas sobre diferentes tipos de dados e efeitos de contenção, causados por concorrência. As estimativas podem ser geradas em poucos segundos e permitem analisar o impacto causado na aplicação pela utilização de diferentes configurações do sistema de processamento paralelo. A ferramenta foi implementada para um sistema de processamento paralelo baseado em Java, chamado JoiN, e ela mostrou ser possível obter estimativas satisfatórias de tempos de execução para diversos tipos de aplicações paralelas / Abstract: Performance prediction is an important mechanism to evaluate the use of resources and predict the execution time of applications in parallel systems. This work presents a tool for the creation of models that represent applications and computers characteristics. Such models are combined to generate a performance model, whose analysis derives execution time estimates that include delay factors, such as arithmetic operations on different data types and contention effects caused by process concurrence. The estimates can be generated in a few seconds and allow the analysis of the impact caused in the application by changes in the parallel system configuration. The tool was implemented for a Java based parallel processing system called JoiN and it showed that satisfactory execution time estimates can be obtained for several types of parallel applications / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

In this Master Thesis, several common applications and Windows operating systemservices are modeled within the field of information security. This thesis focuses onapplying the Enterprise Architecture Analysis Tool (EAAT) and the Cyber SecurityModeling Language (CySeMoL), which are developed by the Department of IndustrialInformation and Control System (ICS) at KTH. The overall objective of this study is todetermine the probability of the CySeMol model with a particular kind of attack. Theproject models six common applications on Windows platform and two Windowsoperating system services. The detailed information regarding the applications and defensemechanism are acquired from various sources. A few experiments have been carried out tovalidate the correctness of the predicted probabilities calculated by the CySeMoL. Theresults of the analysis suggest that the CySeMoL model has a good performance onoperating system vulnerability prediction. At last, some possible suggestions in the contextof the CySeMoL model are given.

In this Master Thesis, several common applications used with MySQL and Linux server are modelled using the Enterprise Architecture Analysis Tool (EAAT) and the Cyber Security Modelling Language (CySeMoL), both developed by the Department of Industrial Information and Control System (ICS) at KTH. The objective of this study is to use the CySeMoL tool to evaluate the feasibility and correctness of the tool by simulating some particular type of attacks on a real life Linux server. A few common applications with MySQL on a Linux server and two Linux operating system services are modelled and explained together with their detailed information and defense mechanisms. A real life penetration test has then been carried out in order to validate the simulated results from the tool. The results of the analysis suggest that the security vulnerability predictions done by CySeMoL on a Linux server has good predictive performance. / I denna Masteruppsats modelleras ett antal vanliga applikationer på en MySQL- och Linuxplattform med hjälp av Enterprise Architecture Analysis Tool (EAAT) tillsammans med Cybersecurity Modeling Language (CySeMoL). Båda dessa är utvecklade vid avdelningen för industriella informations- och styrsystem (ICS) på KTH. Syftet med denna studie är att validera korrektheten av CySeMoL-verktygets sårbarhetsprediktioner genom att simulera ett antal specifika cyberattacker mot en Linuxplattform. Ett antal vanligt förekommande applikationer på en MySQL-plattform samt två operativsystemstjänster i en Linuxserver modelleras. Penetrationstest utförs därefter för att validera resultaten som simuleras i CySeMoL-verktyget. Studien visar att CySeMols förutsägelser stämmer väl med resultaten av penetrationstesterna.