Metoda Data Vault a její implementace při budování datového tržiště

The topic of this diploma thesis is the Data Vault method, which is a specific, hybrid approach to data warehouse modelling in between classic normalized and dimensional modelling, and amidst other benefits enables historization of data and data source tracking. The core purpose was to discuss the Data Vault method specifics and implementation benefits and illustrate the method on real-life project use.The main aim of this paper is to illustrate the use of the Data Vault method for a development of a data mart. In the first part of this paper the Data Vault is introduced, along with its birth, theoretical background, specifics and standards. The Data Vault method is compared with classical, dimensional principles of data warehousing and its benefits and negatives are described.Secondly, the Data Vault method is applied within the real-life project of Claims and Complaints data mart development for an unspecified logistics company. Besides general information about the project there is a focus on the creation of Loading Patterns – patterns for the data load of hubs, links and satellites. They are identified, implemented for the application in SQL load script development and the future possibility of automated code generation is discussed. The developed loading scripts is compared to their patterns and its functionality is checked.Another, secondary focus point is the problematics of connecting standard reporting tools to the data mart via proxy dimensional layer.The last part contains discussion of the problems that have surfaced during the project solution and the lessons learned, which will be later used in the projects planned for future.