For real time decision making in any enterprise, the need of the hour is information gathering. The challenge is to gather information across multiple business units and analyze for further actions. Due to legacy and local business unit level decisions, disparate systems with heterogeneous technologies, platforms, databases, schemas exist. Different business functions are having packaged or custom implementations. These implementations most of the time are based on the application specific requirements not the enterprise big picture in mind. As these practices are being continued in a uncontrollable fashion, the big picture gets more redundant dots. Think of enterprise blue print with different packages as boxes. If one starts to color the boxes which are similar in function in the same color, current enterprise blue print will have categories of redundancies in multiple colors. One may wonder that this whole exercise has no ROI because of the huge cleaning up cost involved. Caveat is the peter's principle of programming. "Complexity tends to increase until it reaches a point where it can no longer be managed, creating an uncomfortable equilibrium" Instead of sitting tight in the discomfort zone of unmanageable equilibrium, start analyzing the state of a enterprise using Zachmann?s Framework . In this context, specific to Data Dimension in Zachmann Matrix.
Moving on the cells of Zachmann Matrix from Data to Data Definition. The metadata repositories are prevalent in enterprises to sync up the enterprise data dictionary. This is happening in places where enterprise architecture group exists and buy in is there from all stake holders. Physical and Logical Data Model Thanks to database modeling tools, the physical and logical data models are captured at the design stage. The challenge continues how to refactor the redundancies in multiple schemas either due to packaged implementations or in the absence of any process of modifying the database schemas. Extension of attributes and the metadata management in the packaged implementations make life tougher.
Transaction, Reporting, Analytical models are being identified at the business model level. Entities, relationships, different dimensions and the facts are designed and maintained for Business intelligent packages or custom analytical engines. Custom ETL are designed to flatten dynamic attributes and stage the data for DataMarts. The latest reporting tools are capable of scheduling of report processing and the capability of looking at the last processed report .
Staged databases which has collated data from disparate sources are being mined for information. The mining process is typically associated with a knowledge base which is used for classification and clustering. Knowledge base implicitly has the learning incorporated in it.
Though the emphasis is put on Zachmann?s framework in this article, enterprise architect have a choice of picking any methodology which makes sense for their enterprise. The need for any enterprise is to do a gap analysis between current architecture and target architecture. Planning needs to happen to bridge the gap in small steps. Before the execution, it is very important to get the buy in from all the stake holders in the methodology and the value in creating a true enterprise architecture.( Dec 27 2005, 09:30:45 PM EST ) Permalink Comments 
This is a personal weblog, I do not speak for my employer.
|« March 2006