University of Twente Student Theses

Login

The design of a Delta Impact analysis model for data warehouses

Kok, J. (2007) The design of a Delta Impact analysis model for data warehouses.

[img] PDF
1MB
Abstract:Motivation The field of data warehousing has emerged over the last decades. A data warehouse is developed at a moment in time to support business intelligence for an indefinite period of time. During the life of the data warehouse the world around it evolves, including the systems that are a source for the data warehouse. In order for a data warehouse to remain functioning and guarantee the quality of its data, it needs to be adjusted to the evolving world around it. The concept of Delta Impact Analysis is used by the company BI4U for the activities of analysing the impact of specific changes. This concept is important because it provides insight into how a data warehouse can be adjusted to the evolving world around it. The motivation to perform this research was the fact that a clear definition on the concept of DIA and what it comprehends was lacking. Goals The main goals of the research were to examine the topic of DIA in practice, to gather insights from literature and other research, to design and develop a practical model for DIA, to test the DIA model, and to provide recommendations to better support changes in data warehouse source systems. These goals resulted in the following main problem statement: How can a Delta Impact Analysis model be designed that supports the process of analyzing the impact of changes in a data warehouse source system situation? Approach The research approach is based on the design science framework by Hevner in combination with action science theory to validate the resulting DIA model from the research. The design science perspective resulted in an approach that is both rigorous, by performing a literature study, and relevant, by applying the research to practice. The research started by investigating the concept of DIA in practice at BI4U, in order to provide more insight into what it comprehends and what was relevant for the focus of the research. Next a thorough literature study was performed. Finally an artifact was proposed, a model for the process of DIA, which was validated in practice with a field study. Results The research offers several contributions. The first contribution is the fact that it was established that several aspects of DIA were unclear at BI4U. No clear guidelines existed with regard to what tasks exactly were part of DIA, and how these should be performed. Also no clear framework was present concerning which types of changes could occur and what kind impact these could have. However, a better vision on the scope of DIA was established, which provided a point of focus for the literature study. The second contribution is the fact that in the literature study insights from the field of impact analysis were applied to that of data warehousing. Categorizations are offered to classify delta and impact. The delta can be distinguished by its origin: schema, semantic or business rule. A schema delta concerns structural changes to a source database, a semantic delta concerns changes to the meaning or representation of data in the source, and a business rule delta concerns changes to the meaning of source data for the data warehouse. The impact can be distinguished as: information-preserving or information-changing impact on the information supply to the data warehouse. The ‘changing’ type can be divided by an information supply that has been reduced, increased and redefined. For the different delta categories it is specified what impact categories can apply, this provides a bridge between defining a delta and determining the impact in impact analysis. To deal with the different types of information-changing impact one has the choice if the core of the data warehouse is not modified, modified and/or modified with regard to historic data. The final and most significant contribution of this research is that the results of the first two contributions are combined and applied in an artifact. This artifact is a model for the process of Delta Impact Analysis. The model describes seven main steps that should be performed to execute a solid DIA, from identifying the concrete change and impact till a proposed solution including required resources and costs, communication of this and an evaluation. Evaluation criteria for the process were determined and used for a practical field study to validate the proposed model of this research. The field study provided insight into which aspects can be improved in the future. Conclusions and discussion The result of this research, the model for the process of Delta Impact Analysis, provides a solid starting point for tackling source system change situations for BI4U, and also for others in the field of data warehousing. Several recommendations can be given with regard to practice. First of all the model should be improved by consistently using it to execute Delta Impact Analyses and adjusting it according to the outcome of the evaluation of the process. Also it will be interesting to perform future research on the usability of existing tools and/or the development of a tool to support the steps in the DIA process. With regard to the scientific field several recommendations can be given for future research. For instance more research could be done concerning which resource and cost estimation models and techniques could be used for DIA. Also it will be interesting to research how the DIA model can be expanded to a model that can be applied to tackle any type of change in implemented data warehouse solutions. The topics of data warehouse evolution and meta warehouse were identified as topics of interest that can support the process of DIA.
Item Type:Essay (Master)
Clients:
BI4U (Business Intelligence for You)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business Information Technology MSc (60025)
Link to this item:https://purl.utwente.nl/essays/58155
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page