Patterns

Patterns of Enterprise Application Architecture

Fowler's POSAA catalog, organised by chapter.

Patterns of Enterprise Application Architecture is Martin Fowler’s catalog of the recurring structures in enterprise software: layering, domain logic, object-relational mapping, concurrency, session state, distribution, and the base patterns they build on. The catalog below is organised by Fowler’s chapter categories.

Domain Logic Patterns

How business rules are organised in code — from Transaction Script’s one-procedure-per-action to Domain Model’s object graph, with Table Module and Service Layer as the intermediate and boundary forms.

Data Source Architectural Patterns

How application code talks to a database: Table Data Gateway and Row Data Gateway expose the schema, Active Record binds a row to a domain object, and Data Mapper keeps domain and schema independent through a separate mapping layer.

Object-Relational Behavioral Patterns

The behavioural machinery of mapping: Unit of Work tracks what changed in a transaction, Identity Map keeps one in-memory copy per row, and Lazy Load defers fetching associated data until it is needed.

Object-Relational Structural Patterns

How objects and relational structure line up: Identity Field, Foreign Key Mapping, Association Table Mapping, and Dependent Mapping for relationships, plus Embedded Value, Serialized LOB, and the three inheritance-mapping strategies for the awkward parts.

Object-Relational Metadata Mapping Patterns

Putting the mapping itself in metadata rather than hand-written code: Metadata Mapping drives the mapping from a description, Query Object builds queries as objects, and Repository mediates between domain and data layers with a collection-like interface.

Web Presentation Patterns

Separating HTTP handling from presentation and domain logic: Model View Controller, Page Controller and Front Controller for request routing, Template View and Transform View for rendering, and Application Controller for dispatching.

Distribution Patterns

Making remote calls affordable: Remote Facade wraps fine-grained objects in a coarse-grained interface, and Data Transfer Object carries data across the boundary in a single round trip.

Offline Concurrency Patterns

Concurrency when a business transaction spans more than one system transaction, so several users can hold a copy of the same record before anyone saves. Fowler’s four patterns cover the two ends of the tradeoff, plus how the lock is acquired and how far it reaches:

Base Patterns

The small, widely reused building blocks the rest of the catalog leans on: Gateway, Mapper, Layer Supertype, Separated Interface, Registry, Value Object, Money, Special Case, Plugin, Service Stub, and Record Set.