RESTful API Patterns & Practices • Mike Amundsen • GOTO 2024

Pattern thinking helps create reusable, composable software solutions by identifying recurring problems and their core solutions

Data should be viewed as evidence of business capabilities rather than the primary focus - actions and workflows are what matter

Client applications are more important than services - services enable clients to do work, but clients make the actual decisions

APIs need to be forward-compatible and handle changes over time through:

• Versioning
• Extensible data models
• Hiding implementation details
• Supporting multiple formats

Messages should focus on:

• Actions/capabilities vs raw data
• Common vocabularies and semantics
• Metadata for queries and responses
• Identifiers for all elements
• Supporting multiple formats (JSON, XML, etc)

Key service patterns include:

• Service discovery and health monitoring
• Consistent error handling
• Caching and optimization
• Stateful workflow support
• Idempotent operations

Workflow needs to support:

• Step-by-step processes
• Rollbacks and reversibility
• Client state management
• Service orchestration
• Error recovery

Systems should be designed for:

• Uncorrelated parties to interact
• Unknown future use cases
• Changes in underlying technology
• Long-term evolution
• Global scale

The data model is not:

• The object model
• The resource model
• The representation model

Focus on making designs:

• Composable
• Extensible
• Modifiable
• Reversible
• Resilient