Enterprise Logical Data Model

At the foundation of a successful Business Intelligence implementation is a Enterprise Logical Data Model.  A Logical Data Model (LDM) defines business requirements in business terms.  It defines the relationships and interactions between organizations and customers and suppliers.  These definitions become the origination for every layer within a typical business intelligence environment.  You can think of the LDM as the glue that gives the BI environment the “single version of the truth.”  From the LDM, Physical Data Models and Metadata Models are derived.

Physical Data Models

The Physical Data Models transforms the logical data model into a database design.  Within business intelligence there are typically two types of data models: 1) Enterprise Integration Data Models and 2) Business Application Data Models. 

Enterprise Integration Databases are typically come in the form of an Enterprise Data Warehouse or sometimes called a Hub.  This type of database is data centric with the logical data model as its only set of requirements.  The objective of this database is to integrate enterprise data into a single language — a physical representation of the logical model.

Business / Application Databases are typically in the form of a data mart or application database.  These databases present the data in the Enterprise Integration database in a form that can be applied for a particular business application or analysis.

Metadata Models

The Enterprise Logical Data Model helps provide context for the organization of metadata.  Metadata is data about data.  More specifically provides information about:

• how data is created
• the purpose of the data
• time and date of creation
• creator or author of data
• source of data
• what standards were deployed

There are many applications of metadata.  In relation to enterprise information, we will focus on 3 areas: Integration, Business Intelligence and Unstructured data.  The Enterprise Logical Data Model provides the linkage and continuity between these several sources.  The Enterprise Logical Model provides the common language for business concepts and processes.

Integration Metadata Models represent data mappings between source and target systems.  Included in this layer of metadata are transformation and logic rules.   This metadata is often collected as part of the integration requirements gathering and is used as part of the integration design.  The metadata will also persist as part of an ETL tool’s metadata.  Beyond development, Integration Metadata is useful for impact analysis.   When the metadata is complete and current, any changes to the data structures can be traced to dependent systems and is a great asset within the change management process.

Business Intelligence Metadata Models provide the mappings between physical data structures and business concepts.  It is within this context where data is transformed into information through facts, attributes, calculations and filters.  It is through the Business Intelligence Metadata that most users interact with the data through reports and analytics.  Without the Business Intelligence Metadata, users would have to know where data was located and how it was structured.

Unstructured Metadata Models organize the unstructured world of the web sites, document repositories, media libraries, etc.  By using the structure of the Enterprise Logical Data Model, these various forms of electronic media can be mapped to the structured data within an organization and provide additional insights.  For example, think of a call center.  The log of a particular call can be indexed in such a way that it is linked to the store for which a complaint is filed.   This could then be used as part of other analytic reporting such as sales reporting to provide anecdotal insight to problems that might be impacting sales for a given day.