Research And Implementation Of Real-time Streaming Updates And Materialized Sample View In Dynamic Data Warehouse

As a valuable resource in modern enterprise, information has been playing an increasingly important role. Data warehouse, as a data carrier, is designed to support decision-making effectively by comprehensive information management. Traditional data warehouse usually focus on internal strategic decision-making, while dynamic data warehouse mainly focus on enterprise tactical decision-making. Dynamic data warehouse is a conceptual innovation, but its technical architecture is still based on conventional data warehouse.ETL (extract, transform, load) as a very important part within data warehouse building, is playing a key role throughout the whole project. Dynamic data warehouse requires more in real-time ETL, so that the real-time ETL is becoming a hot spot in data warehouse research. And as another important part in data warehouse, materialized view is designed to improve query response and reduce the calculation workload. So the design, selection and maintenance of materialized view are other hot spots in data warehouse research. This dissertation will take the data warehouse used for a large financial information system in a multinational as background, discuss and analysis many aspects of issues within enterprise dynamic data warehouse building and implementation. It will also introduce data warehouse architecture and each component, and research for the crucial technologies systematically.The dissertation research and analysis a key technology in real-time ETL, known as real-time streaming updates, advanced a D-MESHJOIN algorithm for distributed dynamic data warehouse, and also proved its good performance and expansibility in distributed environment.Moreover, the dissertation also research on a special materialized view, called materialized sample view, design a new view structure based on ACE+tree. This type of view is upgraded from ACE tree, resolved the issue that sample view can not support dynamic incremental maintenance. Furthermore, the dissertation also gives a query strategy for distributed materialized sample view. And the effects of all the above work have been proved in experiments.With the research and development of dynamic data warehouse, the dissertation gives the solutions to corresponding problems. And it will be helpful and valuable for similar enterprise dynamic data warehouse building, real-time ETL and materialized sample view implementation.