What is structure and design?

The structure and design discipline focuses primarily on data modeling. The goal is to provide direction on standards, best practices, and the use of conceptual, logical, and physical models, naming standards, and tools.

What is a data model?

A data model is a graphical means of documenting data requirements and defining database design specifications. It is used by data modelers to:

• Confirm understanding of data requirements with business owners; and
• Provide direction to technical teams who must ultimately built the database structures.

What types of models are required?

The following models are required to support information management:

• Conceptual model, is generally created at the information management strategy stage.  It contains key entities and relationships and presents a high level look at all of the entities within an organization.

• Enterprise data model is an entity relationship diagram which builds on the conceptual data model and adds additional details;

• Logical model, is a fully attributed entity relationship diagram (ERD), which shows each entity, its relationship to other entitles and specifies the applicable business rules;

• Dimensional Model, represents facts and dimensions, which are used for reporting and analytical purposes. There is no logical representation of a dimensional model. It is a “physical” model but sometimes it is shown in a logical version in  the modeling tool with English like names as opposed to physical names; and

• Physical Model, is the final representation of the data base structures that will be generated from the model. It contains the detailed specifications for the database design and, in a model driven environment, the modeling tool will generate the data definition language (DDL) that is used to create the database structure.

When are data models created?

• The conceptual model is created at the information management strategy stage;

• the enterprise data model is created as part of the information management framework;
  

• The logical model, and conceptual dimensional model are created during the requirements analysis phase; and

• A physical model and dimensional model are created during the architecture and design phase.

Do we always need a logical data model?

A logical data model is a mandatory requirement for an on-line transaction processing system. It is required to show the correct business rules that need to be applied to the information.

A logical model may not be required if the project only involves information movement.  These types of information management projects might move data from several sources and consolidate it into one target system.

Although we may have interim data storage, it is transient in nature and sometimes there is little to be gained by creating a logical model.


The following should be addressed in structure and design standards and best practices.

An enterprise conceptual model should be created as part of the information management strategy. This model is part of the business model and shows what classes of information are required to support management and business intelligence purposes.

An enterprise logical model (EDM) should be created as part of the information management framework. This model should eventually contain all entities and their relationships and a complete set of documentation.

An application logical model should be created during project requirements analysis phase:

• This should be based on the enterprise conceptual model created in the information management strategy phase; and
• This model should show what information is required to support the specific application. If new entities or definitions are discovered during project requirements analysis, they should be added to the EDM; 

Application physical data models should be created during the project architecture and design phase and should be:

• Based on the logical model;
• Created for each environment that will be used by the application; and
• Used to generate data definition language (DDL) for each database

Environment models should be created for each data base environment; and
The environment model should generate the data definition language (DDL) used to create database schema's and objects.

A physical model should specify the data base design and should always be synchronized 100% with the database

Standards for data modeling tools should be established.

Naming standards for the enterprise data model should be included in the data model standard.
Data warehouse data modeling techniques should be introduced to all data modelers.
Standards for data warehouse dimension models must be established.
The design plan for data warehouse should be considered as part of structure and design.
Rapid data warehouse design techniques should be explored to determine what is practical for the organization.
Data warehouse data model standards should be defined.
Data warehouse design solutions should be based on the enterprise data model.
A fully attributed entity relationship diagram (ERD) should form the basis for the EDM.
A conceptual data model should be completed as part of the information management strategy study.
All data modelers should be aware of data modeling concepts
Data warehouse models should be based on the EDM.
Logical data models should be completed as part of the project requirements analysis phase.
Data warehouse designs should be generated from physical data models.
Data modeling naming standards should be included in the data model standard.
Data warehouse models should be based on the EDM
A data modeler should be involved in the requirements analysis phase to create logical data models.
Position descriptions for data modeler jobs should include information management specific roles and responsibilities.
On-going data modeling training should be available for prospective data modeling candidates.
A business model and conceptual data model should be created as part of the information management strategy.
Dimensional model naming standards should be included in the data model standard.

Summary...

An enterprise data model is a key component of an information management framework. Standards and best practices, and fully trained staff,  are required to ensure rapid project delivery and optimal return on information management investment.