Data Architecture from an Enterprise Architecture Perspective

by Jeff Tash, ITscout

TOGAF (The Open Group Architecture Framework) defines four types of architecture that are subsets of an overall Enterprise Architecture:

1. Business Architecture
2. Data Architecture
3. Application Architecture
4. Technology Architecture

Business Architecture describes business processes.  Data Architecture, this article’s topic, defines business objects.  Application Architecture specifies business solutions which are comprised of a ‘sprinkle of process’ and a ‘dash of data’.  Technology Architecture organizes and describes an enterprise’s technology portfolio.

The goal of architecture is to document everything. Formally codifying models provides a framework for capturing and organizing content. Modeling and documenting is nearly worthless unless the captured and organized information can also be easily communicated to the people who need to access it. 

What is Data Architecture? 
Data Architecture is a model mapping to critical business entities as well as relationships among entities.  Each entity includes descriptions of properties and behaviors.  Encapsulation hides implementation details about data, such as the distinction between fields and operations (i.e., methods).

Data Architecture represents models that reflect relationships among instances of real-world entities.  These include one-to-one, one-to-many, and many-to-many relationships.  Support is also needed for metadata relationships, especially inheritance which distinguishes between generic cases versus special cases.  Composition is another important relationship used to describe things made up of other things.

Who needs access to Data Architecture? 
Start by defining personas that describe people who care about data.Personas are not real people, but hypothetical archetypes of actual users.  Personas are defined by goals. Goals are defined by personas. 

One important persona is the business executive who needs Data Architecture to better understand his or her own business environment.  Literally, business people need a cognitive roadmap to help them better communicate and understand technology. 

Another important persona involves business analysts who require access to Data Warehouses and/or Data Marts.  These folks need to view information in terms of subject databases that are organized using multi-dimensional attributes that can be sliced & diced or rolled-up & drilled-down. 

Software developers need to understand Data Architecture from the perspective of either directly accessing data or invoking methods that indirectly access data.  Developer productivity is greatly enhanced by the automated support of declarative constraints and validity tests, such as referential integrity.

The effectiveness of Data Architecture is measured by a combination of its accuracy in modeling real-world entities and its accessibility by a widely diverse audience.  

The beauty of Data Architecture is its innate stability -- especially by comparison to Business Architecture which forever is in a continual state of flux.  Data Architecture models real-world things, like customers, parts, suppliers, contracts, employees, etc.  These things don’t tend to change very often.  Enterprises can reorganize and reengineer all they want.  At the end of the day, a customer is still a customer; an employee is still and employee; a supplier is still a supplier; etc. 

The job of the Data Architect is to define for a single piece of data multiple viewpoints to support different personas.  Obviously, there must always be consistency across all viewpoints, but different audiences can be interested in different facets of the same data.  For instance, IT people with operational responsibility for backing up and restoring data have very different perspectives on data than software developers or business analysts.  The same holds true for security managers.  

Conclusions
Data Architecture is one of four core pillars that constitute Enterprise Architecture. 

At its core, Data Architecture is a model of those entities -- objects -- things -- that are important to the enterprise’s real-world environment.

Like a diamond, Data Architecture is multi-faceted with different audiences interested in different characteristics of data. 

The quality of an enterprise’s Data Architecture is a great predictor of an IT organization’s overall success.