Submitted by Luiz Felipe Barbosa (luiz.fbabreu2@ufpe.br) on 2015-03-12T15:00:05Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação Samara Nascimento.pdf: 9257843 bytes, checksum: 902eb4bdc806735989a42837a7f7bbe3 (MD5) / Approved for entry into archive by Daniella Sodre (daniella.sodre@ufpe.br) on 2015-03-13T13:15:07Z (GMT) No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação Samara Nascimento.pdf: 9257843 bytes, checksum: 902eb4bdc806735989a42837a7f7bbe3 (MD5) / Made available in DSpace on 2015-03-13T13:15:07Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação Samara Nascimento.pdf: 9257843 bytes, checksum: 902eb4bdc806735989a42837a7f7bbe3 (MD5) Previous issue date: 2013-03-25 / CNPQ / A técnica experimental de avaliação de desempenho utilizada em aplicações e sistemas de bancos de dados é composta principalmente da técnica de benchmark, que consiste em um conjunto de testes experimentais previamente definidos e posteriormente executados para obtenção de resultados de desempenho. Data Warehouses Geográficos (DWG) permitem o armazenamento de geometrias dos objetos que representam localizações na superfície terrestre e possibilitam o processamento de consultas analíticas e multidimensionais. Os benchmarks TPC-D, TPC-H e SSB são utilizados para avaliar o desempenho de Data Warehouses Convencionais. O benchmark Spadawan é utilizado para avaliar o desempenho de Data Warehouses Geográficos. Contudo, os benchmarks anteriores não conseguem ser considerados abrangentes, devido a sua limitada carga de trabalho. Desta forma, nesta dissertação, propomos um novo benchmark, chamado Spatial Star Schema Benchmark, ou Spatial SSB, projetado especialmente para realizar a avaliação de desempenho de consultas em ambientes de DWG. As principais contribuições do Spatial SSB estão concentradas em três pontos. Primeiro, o Spatial SSB utiliza três tipos de dados geométricos (i.e. pontos, linhas e polígonos), propostos em um esquema híbrido. Além disto, garante o controle da seletividade, que indica o número de linhas retornadas na tabela de fatos para cada consulta espacial pertencente à carga de trabalho deste benchmark. Segundo, o Spatial SSB controla a geração e distribuição dos dados no extent, assim como a variação do volume de dados, tanto aumentando a complexidade dos objetos espaciais, quanto aumentando o número de objetos espaciais, pelo aumento do fator de escala. Terceiro, o Spatial SSB obtém o número de objetos intersectados por janelas de consultas definidas de forma ad hoc, que sobrepõem uma porcentagem do extent definida pelo usuário. Os resultados experimentais mostraram que estas características degradam significativamente o desempenho de consultas sobre DWG.

Made available in DSpace on 2014-06-12T15:59:30Z (GMT). No. of bitstreams: 2 arquivo5134_1.pdf: 2541591 bytes, checksum: 6f46eb970bb56cef73fafe13dc208cad (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2002 / O volume de informações a ser trabalhado na tomada das decisões gerenciais supera largamente a capacidade do processamento humano, mecânico e dos sistemas transacionais atuais, exigindo ferramentas de apoio à decisão mais adequadas aos novos desafios gerenciais. Mesmo aplicando-se modelos de decisão tidos como adequados, uma grande parte das implementações de Sistemas de Informação não atingem os resultados esperados, o que levam muitos deles ao fracasso total ou parcial. Acredita-se que com obtenção de resultados rápidos se possa conseguir um maior envolvimento do usuário final, o que segundo os especialistas diminui bastante a possibilidade de fracasso. Esse trabalho visa a utilizar técnicas de análise e exploração de dados na construção de soluções de Sistemas de Apoio à Decisão, em especial na construção de Data Warehouse(DW). Aproveita-se o conhecimento adquirido com a aplicação dessas técnicas, mostrando a sua importância nas diversas fases de sua construção de um DW. Propõe-se e implementa-se uma metodologia chamada FASTCUBE, que é baseada em um modelo de préprocessamento de dados. Ela incorpora de maneira rápida os metadados extraídos diretamente da massa de dados. Acelerar e sedimentar a compreensão do problema, sempre levando-se em consideração a qualidade dos dados, durante todas as suas fases é um dos pontos forte dessa metodologia. O seu objetivo final é acelerar o processo de visualização do modelo de decisão, através de um protótipo de modelo dimensional, com dados operacionais amostrados no início do processo e tratados durante o mesmo

Nowadays, business intelligence techniques are applied more and more often in different settings including corporations and organizations both in the private and public sector. It is really a broad field which can assist business people to realize the state of their organization and make profitable decisions. In this thesis, I will focus on one of its components, data warehouse, by proposing activity theory as the method to solve the dimension identification problem in data warehouse. Under the background of project IMIS and the involved personnel, who determine the dimension, firstly I study how to use the ER method, “bottom up” method, and activity theory method to identify the dimension in data warehouse, and some relevant knowledge about the three methods. Then, we apply the three methods to identify the dimension. After that, I evaluate the dimension identification results of the three methods according to the feedback from the healthcare organization to get their veracity and integrality. Finally, based on the results of my efforts, I arrive to the conclusion that the activity theory method can be applied to identify the dimension in data warehouse, and with the comparison to the other two traditional methods (ER model and “bottom up”), the activity theory method is more easy and natural to identify the dimension of a dimensional model.

öretag har under lång tid använt sig av olika typer av beslutsstödjande system för att fatta rätt beslut. Under 1990-talet har en typ av beslutsstödjande system som kallas datalager utvecklats. Datalagret har utvecklats till ett beslutsstödjande system som hjälper slutanvändarna att analysera data med hjälp av olika typer av verktyg med användarvänliga gränssnitt. I rapporten undersöks hur datalager används av företag i deras verksamhet. Som underlag för rapporten har fem företag studerats. Bland företagen som ingick i undersökningen blev svaret att de använde sina datalager främst till att ta fram statistik och genomföra uppföljningar. Företagen visade även upp ett antal företagsspecifika anpassningar av sina datalager. De företag som ingick i undersökningen visade även på en bred spridning av användandet av datalagren inom företagen. Antalet användare på företagen varierade från 20 st till 200 st. Undersökningen visade även att spridningen av användandet hierarkiskt i organisationen generellt var väl utvecklat.

I datalager samlas information från flera olika system och sparas på ett gemensamt sätt. Informationen i ett datalager ger en holistiskbild över organisation och bör användas som beslutsstöd. För att utvinna information ur datalager finns flera olika metoder och informationen kan sedan användas inom flera olika områden. Syftet med studien är att undersöka vilka metoder som används för att utvinna informationen ur datalager inom industribranschen samt inom vilka användningsområden informationen används. För att besvara frågeställningen genomfördes fem intervjuer med företag i industribranschen. Resultatet visar på att vissa metoder för att utvinna information ur datalager används framför andra och inom datautvinning används inte den förutsägandekategorin alls. Framförallt använde sig företagen av informationen på kundsidan och såg det som ett levande projekt som skall täcka in hela processen i framtiden.

För att flytta information från en databas till ett datalager används det idag olika tekniker. Existerande transformeringstekniker baseras på att en applikation hanterar detta. Detta examensarbete går ut på att skapa och undersöka en metod som istället genomför transformeringen i en databas. Denna transformering är metadatadriven, eftersom metadata är den information om data som krävs för att en transformering ska vara möjlig. Arbetet bygger därför på en metadatastudie som behandlar representation och struktur av metadata. Målet med arbetet är att få fram en så generell transformeringsmetod som möjligt och metoden går ut på att transformera data från en normaliserad databasstruktur till en denormaliserad datalagersstruktur.

This diploma thesis deals with building of large Data Warehouses on Microsoft SQL Server 2008 platform. First part of the thesis discuss about news in MS SQL Server 2008, which I have considered as being important for Business Intelligence area. Following chapters are focused on approach to implementation of real Data Warehouse for Customs Administration of Czech Republic on MS SQL Server 2008 basis. These parts of the thesis cover specifics of the Customs Administration of Czech Republic, current state of Data Warehouse and reasons for migrating to MS SQL Server 2008. Further the thesis describes both logical and physical architecture of proposed solution and a way of implementation Data Warehouse on MS SQL Server 2008. The main aim of the thesis is to create a list of critical success factors (CSF) of building large Data Warehouses on MS SQL Server 2008 platform. It's not only plain list CSF, but rather the best practices for MS SQL Server Data Warehouses implementation. The most significant contribution of this thesis is that it offers instruction manual for designing and implementing large Data Warehouse on MS SQL Server basis. Background for the creation of summary of CSV and recommendations were vendor's publications and especially my own experience with the product. My colleagues from company Adastra gave me an advisory opinion as well.

The Teradata database system is specially designed for data warehousing environment. This thesis explores the use of Teradata in this environment and describes its characteristics and potential areas for optimization. The theoretical part is tended to be a user study material and it shows the main principles Teradata system operation and describes factors significantly affecting system performance. Following sections are based on previously acquired information which is used for analysis and optimization in several areas of described environment.

This dissertation titled Designing Project Business Intelligence focuses on data warehouse in a medium-sized company. The theoretical portion of the thesis describes the topic of data warehouse and dimensional modeling. In the first theoretical chapter, I compare and contrast two approaches to datawarehouse building - Bill Inmon's and Ralph Kimball's. Then, basic terminology used in the field, principles of dimensional modeling and various approaches of database use will be described. Furthermore, the thesis illustrates the role of the Mondrian system as a mediator between the relational database and the OLAP server. Possible uses of the MDX language in working with multidimensional databases are outlined. The theoretical portion of the thesis ends with a description of the applicational interface XMLA, which can be used for facilitation communication between applications and the OLAP server. The practice-oriented portion of the thesis includes a complete design of the Business Intelligence project. The project is divided into the following parts: Feasibility Study, Project Planing, Business Requirements, Design and Development, and Deployment and Maintenance. The entire project operates exclusively on Open Source Software. Consequently, the project shows that small and medium-sized companies can afford to run a fully operating Business Intelligence system.

The primary objective of this work is to investigate the potential of utilizing Hadoop and Amazon Redshift in the Amazon Web Services ("AWS") cloud, in order to design and implement a data warehouse, the efficacy of which will be tested afterwards. Contributions of this work include: documenting the technologies in the AWS cloud in Czech, demonstration of the design and performance tests of the data warehouse and the ETL part. Another considerable benefit is the added value to the company for whom the project was designed, and which is currently using the output of the project.