1.1.2 Logical Models versus Physical Models
A model is a representation or a simplified copy of something more complex. Architects use models to examine large or complicated structures which are difficult to work with directly. For example, engineers test models of air plane wings or create scale models of buildings.
A data model is a tool which allows the Architect to construct a picture of the data within an organization or a part of an organization. It is not a picture of what the computer sees, but of what the user and business sees.
Logical Data Models are high level or conceptual models which describe data in terms of the user perspective and business rules.
A logical data model represents the business.
A logical data model looks at the basic pieces of information and the inter-dependencies between the information. Logical data models are defined in terms of entities, attributes, and relationships. They reflect the business, its information and rules. It is not designed to represent the Architect's perception of the physical database to be implemented.
A logical data model is platform independent. It is not based on hardware, software, or tool. Logical data models are a description of the business information and not a physical database design.
Physical data models or database designs are low level models which describe data in terms of files and how the information is stored in the computer. When physical modeling, you think of header and detail files, logical views of data, joins of data, etc.. Physical models reflect the technical implementation of the database.
A physical data model represents the computer database.
Physical models look at data files and database design in terms of their implementation. Physical models are defined in terms of fields, record formats, and files. It is a technical representation designed to support the programs and programmer that will use the database.
A physical data model is usually platform dependent. It is designed for a specific computer platform. The hardware, database management system, software language and tool used are all factors in developing a physical model.
As a result, physical models tend to be very inflexible. A physical model cannot be easily moved from one environment to another, or implemented on another hardware platform.