Business Intelligence: Methodological Approach to Data Warehouse

The methodological approach for creating Data Warehouse solutions depends on the company organization, the type of users, the purpose intended, and the system’s technical architecture. To download the full text and to get more information, click HERE

Over time, several approaches have been suggested for project implementation, later revisited thanks to the wealth of experience gained by various Organizations. We therefore speak of top-down approach, bottom-up approach, incremental approach, which correspond to different types of Data Warehouse (Enterprise Data Warehouse, Data Mart, Multi-tier Warehouse). Top-down approachThe top-down approach is one that involves an extensive implementation of the system, whose original design examines from the start all the main areas of business interest. In this case, we talk about an Enterprise Data Warehouse, which can subsequently be divided into a set of Data Marts for technical and organizational reasons. Dependent Data Marts constitute a subset of highly specialized business data for areas of interest or business departments. The weak point of this theoretically rigorous approach lies in the difficulty of managing the all-encompassing project, which risks paralyzing the activity and providing results too far in the future.Bottom-up approachThe bottom-up approach involves an uncoordinated implementation in which each Data Mart is created to meet a specific information need of a departmental user group. In this case, the Enterprise Data Warehouse is the result of the collection of independent individual Data Marts, which feed directly from operational systems. The advantage of such a pragmatic approach is achieving useful results for the user within a limited time frame with relatively contained direct costs. On the other hand, lacking an initial overall vision, the risk is creating segments that cannot be integrated with each other, which result in informational silos that are probably partly redundant and incongruent in their results.“Incremental” approachThe “incremental” approach combines the advantages of the two approaches described above. At the base of this approach, also defined in the literature as a “federated” approach, is the creation of a common information model. From the common information model, coherent data models of the Enterprise Data Warehouse and/or Data Marts are developed; the latter can be either dependent or independent. The implementation foresees pooling the data acquisition processes from source systems among different Data Mart projects. The result of the acquisition processes is centralized in common staging areas where subsequent transformation processes are carried out. The common staging areas are technical areas, not accessible to the end user, used to acquire and process the data that feed both the Enterprise Data Warehouse and the Data Marts. The common information model and the use of staging areas minimize integration problems between Data Marts. The implementation of solutions towards users is faster compared to the top-down approach, because a fully designed enterprise data model a priori is not required, but it is created through an iterative process defining priority thematic areas of interest. Obviously, the centralized management of common documentation requires groupware-type development architectures and somewhat reduces the autonomy (and anarchy) of individual groups. ALL RIGHTS RESERVED

Pubblicato in Business