|
|
Data warehouse based on IBM technologies and its possibilities
Snítil, Jiří ; Pour, Jan (advisor) ; Novotný, Ota (referee)
This diploma thesis deals with the analysis of advanced data warehouse concepts where three advanced data warehouse concepts are analysed and their selection is justified. The first selected advanced data warehouse concept is a method of capturing data changes from sources system Change Data Capture (CDC). The second concept is the historization of captured data into historical data collection. The third concept is the application of analytical functions directly within data warehouse technology. A new testing environment has been created to analyse these concepts where the main database system Netezza available in IBM PureData System for Analytics, powered by Netezza technology (PDA), is utilised. This testing environment allowed all selected advanced data warehouse concepts to be reviewed. An impact of the application of these advanced data warehouse concepts has been analysed based on results from the testing environment and practical insights, particularly regarding potential advances. In the testing environment it was verified that all analysed advanced data warehouse concepts are applicable in a data warehouse. In the first advanced data warehouse concept was chosen LiveAudit mapping as appropriate for further data processing, when with this mapping it is possible to unambiguously determine the state of data in a source system at any point in the past. The second advanced data warehouse concept established that data acquired from LiveAudit mapping is possible to effectively process into historical data collection. Based on these findings, there was proposed generic solution of processing data from source systems. In the third advanced data warehouse concept was also proved, that it is possible to work in native analytic environment RGui and move the computation itself into data, which is located in the data warehouse, without the necessity of migration of these data. Further, it is possible to develop and use a new analytic function written in C++ language directly into the technology of the data warehouse.
|
guest ::
