|
Both approaches are valid and have their benefits, and actually both approaches may be followed at the same time. Developing an Enterprise Conceptual Data Model (ECDM) independent of any specific data design project is very valuable in that it identifies enterprise business objects/concepts and how these relate. The ECDM should be used as a framework for every project touching upon information within the domain/subject area (e.g. Finance, Customer, Sales, Product).
In addition to the ECDM, some organizations also develop an Enterprise Logical Data Model (ELDM), which may be primarily an extension of the ECDM -- the primary difference being that attributes of business concern are added to the entities/objects, or may be relational (e.g. 3NF) in nature (e.g. M:M relationships, subtypes, etc resolved), or may be some combination of both.
Projects may actually develop Conceptual, Logical, and Physical models. At the project level, a CDM may be developed in greater detail than appropriate for the ECDM. A project level CDM can and usually should be used to update/expand the ECDM, especially if what is being modeled is of enterprise concern. ECDMs usually aren't developed for all domains at the same times (definitely not to the same depth), and so it is common for enterprise applications (ERP, Data Warehouse, SOA, etc.) to act as the driver for or be a significant contributor to developing/maintaining enterprise models.
In response to the first part of your question, undertaking a project just to develop enterprise models (Subject Area – identifies domains and very high-level relationships between domains, ECDM, ELDM) is a worthwhile undertaking, but usually has to be associated with an enterprise endeavor to have any hope of funding.
More about conceptual data models
|
