Storing connected data

Relationships are hidden in join tables and foreign keys.

Join tables mix business data with metadata and add complexity.

Foreign key constraints increase development and maintenance effort.

Sparse tables with many nullable columns require extra code checks.

Queries like “friends-of-friends” or recursive queries require multiple expensive joins, which scale poorly with depth.

They use disconnected structures (key-value, documents, columns) without native relationships.

Relationships are mimicked using embedded foreign keys, which breaks down for reciprocal or complex queries.

This approach is not scalable for true graph-like data traversal.

Graph databases store connected data as connected data, directly reflecting real-world relationships.

Suitable for modeling complex domains like social networks.

Flexible: New nodes and relationships can be added without disrupting existing data or requiring migration.

Labels categorize nodes based on their roles (e.g., User, Order, Product).

A node can have multiple labels.

Labels allow efficient querying, e.g., finding all nodes labeled 'User'.

Nodes: Store data as key-value pairs (properties).

Labels: Tag nodes to group them by roles (e.g., User, Product).

Relationships:

• Connect nodes with a direction and a type (no dangling edges).

• Can also have properties for metadata, semantics, and algorithm support