Data Modelling Patterns

Data modelling in DynamoDB is fundamentally different from relational databases. Instead of normalising data and using joins, you denormalise and design your table around your access patterns. This lesson covers the key modelling strategies.

---

The DynamoDB Design Process

In a relational database, you design the schema first and then write queries. In DynamoDB, you reverse the process:

1. List your access patterns — every query your application needs
2. Design your primary key and indexes to support those patterns
3. Model your data to fit the key design

Model	Design Direction
Relational	Schema → Queries
DynamoDB	Access Patterns → Key Design → Schema

---

Single-Table Design

Single-table design is the most powerful DynamoDB modelling technique. Instead of creating separate tables for each entity, you store all entities in a single table using structured partition and sort keys.

Why Single Table?

Benefit	Description
Fewer round trips	Fetch related data in a single Query
Simpler operations	One table to manage, backup, and monitor
Transaction scope	Transactions work across items in the same table
Reduced costs	Fewer tables, fewer indexes

Example: E-Commerce Application

Access patterns:

1. Get a customer's profile
2. Get all orders for a customer
3. Get a specific order with its items
4. Get all items for an order

PK	SK	Attributes
CUST#c001	PROFILE	{ Name: "Alice", Email: "alice@..." }
CUST#c001	ORDER#o001	{ Date: "2024-01-15", Total: 49.99 }
CUST#c001	ORDER#o002	{ Date: "2024-02-20", Total: 29.99 }
ORDER#o001	ITEM#i001	{ Product: "Widget", Qty: 2, Price: 19.99 }
ORDER#o001	ITEM#i002	{ Product: "Gadget", Qty: 1, Price: 10.01 }
ORDER#o002	ITEM#i003	{ Product: "Gizmo", Qty: 3, Price: 9.99 }

Queries:

• Customer profile: PK = "CUST#c001" AND SK = "PROFILE"
• Customer orders: PK = "CUST#c001" AND SK begins_with "ORDER#"
• Order items: PK = "ORDER#o001" AND SK begins_with "ITEM#"
• Everything for a customer: PK = "CUST#c001"

---

Adjacency List Pattern

The adjacency list pattern models many-to-many relationships in a single table:

Example: Social Network (Users and Groups)

PK	SK	Attributes
USER#alice	PROFILE	{ Name: "Alice" }
USER#alice	GROUP#runners	{ JoinedAt: "2024-01" }
USER#alice	GROUP#readers	{ JoinedAt: "2024-02" }
USER#bob	PROFILE	{ Name: "Bob" }
USER#bob	GROUP#runners	{ JoinedAt: "2024-03" }
GROUP#runners	MEMBER#alice	{ Role: "admin" }
GROUP#runners	MEMBER#bob	{ Role: "member" }
GROUP#readers	MEMBER#alice	{ Role: "member" }

Queries:

• Alice's groups: PK = "USER#alice" AND SK begins_with "GROUP#"
• Members of runners: PK = "GROUP#runners" AND SK begins_with "MEMBER#"
• Alice's profile: PK = "USER#alice" AND SK = "PROFILE"