Data Models
How we think about data models ?
Layer 1: As an application developer, we model the real world functional requirements into objects or data structures, and APIs to manipulate those data structures.
Layer 2: These data are then expressed as general-purpose data model, such as JSON or XML documents, tables in a relational database to be stored.
Layer 3: The engineers who built the database then decide on a way of representing the data on disk or on a network. The representation will allow the data to be queried and manipulated in different ways.
Relational Model
This is the model for SQL. Data is organized into relations. A relation (table) is a collection of tuples (rows).
As most application development today is in OOP, for data stored in relational tables, a translation layer is required between the objects in tha application code and the database model of tables, rows and columns.
Document Model
Document Model is appropriate when the data has mostly one-to-many relationships and no relationship between record.
One-to-many and Many-to-many relationship
With SQL database, an one-to-many relationship is usually interpreted with separate tables, and one primary key is used to query all the information. We then perform a multi-way join between the main table and its subordinate tables (like user table and user’s information tables). With a JSON representation, one query is sufficient to ge the tree structure explicit. However, the document representation can make many-to-many relationship difficult.