Chap 02

Database Systems - Chapter 2: Data Models

• Data modeling is an iterative and progressive process for creating a specific data model for a determined problem domain.
• Data models are important for organizing data for various users, acting as a communication tool between designers and end users, and providing an overall view of the database.
• Data models represent data structures, characteristics, relations, constraints, transformations, and other constructs to support a specific problem domain.
• A blueprint for developing a database that satisfies all end-user requirements.
• Over time, several semantic data models have been proposed.
• The relational data model uses a table format.
• The entity-relationship (E-R) model is the most common, often used for communication between database designers and users during database development.
• The E-R model can also be combined with object-oriented models and XML.
• A data model comprises entities, attributes, and relationships/constraints.
• Entities represent a person, thing, or event.
• Attributes are characteristics of an entity (equivalent to fields in file systems).
• Relationships/constraints are rules ensuring data integrity, restrictions on data (e.g., acceptable ranges for values).
• Basic building blocks of a data model are described by business rules.
• Business rules are brief, precise, and unambiguous descriptions of policies, procedures, or principles.
• Business rules describe and distinguish the data’s characteristics. (e.g., 'A customer may generate many invoices' or 'An invoice is generated by only one customer').
• Business rules originate from company managers, policy makers, department managers, and through direct interviews with end-users.
• This will help standardize a company's view of data, acting as a communication tool between users and designers.
• This will allow designers to understand data, role, scope, and business processes to develop appropriate participation rules, constraints, and create an accurate data model.
• Nouns translate into entities in a data model.
• Verbs translate into relationships among entities, using directional relationships (one-to-many, many-to-many, one-to-one).
• Naming conventions are descriptive and facilitate communication and self-documentation (e.g., using descriptive names for entities and attributes).
• Naming conventions are used for entities and attributes. Names should be descriptive using familiar terms for easier understanding. (e.g. STUDENT, STAFF, BOOK, or STUDENT_FNAME, BOOK_ISBN).
• Proper naming conventions in data models are crucial for effective communication and self-documentation.
• Correct identification of relationships for business rules involves establishing relationships between instances of different entities. (e.g., How many instances of B are linked to one A?) and (e.g., How many instances of A are linked to one B?).
• Relational Model: Based on mathematical relations; relations (tables) are two-dimensional, with intersecting rows (tuples) and columns (attributes).
• Relational Database Management System (RDBMS): A system that performs basic functions from hierarchical and network DBMS systems, simplifies data understanding, and hides underlying complexities from users. (e.g. examples: Oracle, Microsoft SQL Server, and MySQL).
• Relationships between tables in a relational model are established through common attributes.

Data Modeling Concepts

• Hierarchical Model: Represents data as an upside-down tree, one-to-many relationships typical, parent segment and child segments (similar to file systems).
  • Advantages: Promotes data sharing, conceptual simplicity of parent-child relationships, database security provided by DBMS, efficient with 1:M relationships.
  • Disadvantages: Requires knowledge of physical data storage, not flexible to change structure, no data definition or standards.
• Network Model: Can represent complex data relationships, improved database performance, schema, and subschema structure.
  • Advantages: Simple concepts, handles multiple relationships, data access flexibility, and conformity to standards.
  • Disadvantages: Requires knowledge of physical data characteristics, not flexible to change data structure, and no standards exist.

Data Models - Evolutions

• File System (1960): Basic structured file.
• Hierarchical & Network (1970s): Data is structured as an upside-down tree with parent/child relationships; parent/child relations are structured in a nested manner.
• Relational (1970s): Data structured in tables with rows and columns.
• Object-Oriented & Relational (1985-1990s): Combines object-oriented features with a relational structure.
• XML-Hybrid DBMS (2000s): XML-based models allow additional flexibility in managing unstructured/semi-structured data.
• Big Data/NoSQL (2009): Newer models to manage vast amounts of data.

Degrees of Abstraction

• Framework for data modeling based on data abstraction: external model, conceptual model, internal model, physical model levels.
• External Model: End users' view of data environment. ER diagrams are often employed. Specific representations of the external view.
• Conceptual Model: Global view of data in the organization. Is independent of hardware and software. Logical design of the conceptual data model.
• Internal Model: Maps conceptual model to the DBMS. Uses chosen database constructs supported by the chosen database. Shows how the database will map to and interact with the DBMS used to physically implement it.
• Physical Model: Describes how data is stored on storage media like disks or tapes, defining the physical storage and data access methods. Is dependent on hardware.

Object-Oriented Data Model (OODM)

• Provides a basis for object-oriented database management systems (OODBMS).
• Structured as objects that contain attributes (properties), relationships, and operations (on objects); organized in a class hierarchy through inheritance.
• Attributes describe the object’s properties.
• Depicted using UML (Unified Modeling Language) class diagrams (used for object hierarchy).

NoSQL

• A new generation of databases designed for massive volumes and unstructured data.
• Offers high scalability, availability, and fault tolerance; supports unstructured data, large volumes of data and speed of data handling.
• Aims to manage large amounts of web and sensor-generated data and provide high performance and scalability at a reasonable cost.
• Can use inexpensive hardware for massive data storage.