Architecture Patterns Catalog

Architecture Patterns Catalog

Curated reference of proven patterns with guidance on when and how to apply them.

Context

You are designing a system and want to know proven patterns that solve your problem. Reference catalog of patterns: when to use, tradeoffs, real-world examples.

Domain Context

Based on pattern catalogs (Gang of Four, Enterprise Patterns, Domain-Driven Design):

• Structural Patterns: How components fit together (Adapter, Facade, Bridge)
• Behavioral Patterns: How components interact (Observer, Strategy, State)
• Architectural Patterns: System-level patterns (Microservices, Event-Driven, CQRS)
• Data Patterns: How data flows (ETL, CDC, Event Sourcing)
• Messaging Patterns: How services communicate (Request-Reply, Pub-Sub, Saga)

Pattern Categories with Examples

1. Communication Patterns:

   • Synchronous (Request-Reply): REST, gRPC. Blocking, tight coupling, simple.
   • Asynchronous (Pub-Sub): Message broker, events. Loose coupling, eventual consistency.
   • Saga Pattern: Distributed transaction across services. Orchestrator vs choreographer.

2. Data Patterns:

   • Event Sourcing: Store events as source of truth. Auditability, temporal queries, replay.
   • CQRS: Separate read and write models. Scalability, complexity.
   • Polyglot Persistence: Different databases for different workloads. Flexibility, operational complexity.

3. Resilience Patterns:

   • Circuit Breaker: Stop calling failing service, fast-fail. Prevent cascade failure.
   • Bulkhead: Isolate resources (threads, connections). One service failure doesn't starve others.
   • Retry with Backoff: Retry with exponential backoff. Transient failures recover.

4. Scalability Patterns:

   • Sharding: Partition data across databases. Horizontal scaling, complexity.
   • Caching: Cache hot data. Latency, consistency trade-off.
   • Load Balancing: Distribute across instances. Horizontal scaling.

Instructions

1. Identify Problem: What are you trying to solve? Scaling? Reliability? Loose coupling?

2. Search Catalog: What patterns address this problem?

3. Evaluate Trade-offs: Benefits vs complexity? Team capability? Do we have the infrastructure?

4. Choose Pattern(s): Often combine patterns. Saga + Circuit Breaker. Sharding + Replication.

5. Implement Carefully: Patterns solve some problems, create others. Monitor carefully.

6. Document: Why you chose this pattern. What trade-offs accepted. What could go wrong.

Anti-Patterns in Pattern Use

• Pattern Memorization Without Understanding: Apply pattern because it's trendy. Result: wrong for problem. Guard: Understand pattern's purpose; ensure problem matches.
• Overusing One Pattern: Everything is microservices or event-driven. Result: wrong tool for job. Guard: Different problems, different patterns.
• Incomplete Implementation: Use circuit breaker but not monitoring. Result: fails silently. Guard: Implement full pattern; test failure scenarios.
• No Evolution Path: Choose pattern, stuck with it forever. Result: outdated, painful. Guard: Design to evolve; refactor when costs exceed benefits.

Further Reading

• Design Patterns: Elements of Reusable Object-Oriented Software by Gang of Four — foundational patterns
• Enterprise Integration Patterns by Gregor Hohpe — messaging and integration patterns
• Building Microservices by Sam Newman — microservice patterns
• Patterns of Enterprise Application Architecture by Martin Fowler — application architecture patterns