The bank created several independent data warehouses and data marts for its Business Intelligence needs over the years, but this approach, despite its unquestionable merits and initial results, led to a variety of conspicuous problems in later stages: data that are never fully integrated, cleaned and consolidated; pieces of data missing in particular warehouses; reports that should be identical appearing differently in various sources; changes in the source system having to be re-analyzed andimplemented in various data warehouses; reporting, already a tedious process in Excel, becoming a highway to hell in terms of slowness and frequency of errors. This state of affairs was screaming for change and Adastra was asked by the client to completely remove these critical shortcomings in their Data Warehousing system.
A consolidated data warehouse, the Analytic Data Store (ADS), was the solution to these problems as it supports data cleaning and data consolidation. It has a separate normalized data storage layer (L1) and a presentation layer for user access (L2).
ADS runs on an Oracle Database. The ADS architecture was completely tailored to the customer's requirements, including integration with the Operational Data Store and the Ataccama Data Quality Center, a tool for advanced cleaning and consolidating data for unification among clients. To clean and consolidate data to the L1 draft layers, we used our own banking data model, a key feature of the solution which once again proved its strength when applied.
For the actual performance data model, we used a high-performance ELT concept which was created using custom-made, advanced Adastra generators using a metadata model in the Sybase PowerDesigner CASE tool to generate a highly effective, efficient and standardized code. The model respects the best practices which we have gained when developing similar solutions in other banks.
The Analytic Data Store was successfully put into production by the bank in 2007 and since then, both the bank and Adastra has operated and developed it. From a user perspective, the ADS has brought about a new and completely integrated and consolidated view of customers, products, accounts, classification, segmentation, approvals and many other areas.
Our client only relies on ADS for the target data warehouse. Thanks to the sophisticated good work with metadata and detailed data flows between the inputs to the data warehouse to the presentation layer, metadata are also used to generate the vast majority of program code which leads to a huge acceleration of the development of newly required changes. In addition to conventional batch processing, ADS also supports near real-time processing for selected jobs. The performance aspect should also be mentioned as it places strong emphasis on parallel processing that enables users to supply reports much earlier than in the older data warehouses. Thanks to the integrated cleaning and consolidation processes ADS reports are characterized by significantly higher quality compared to the original solution.
Following consultations with Adastra, the bank successfully migrated and consolidated its databases on the Oracle Exadata Database Machine platform. Thanks to this, the bank dramatically simplified its IT infrastructure with regard to databases, making them more effective and faster, all while cutting IT infrastructure costs.