Mark Maid ViewerBlog

Event-Driven Architecture: Patterns & Pitfalls

An overview of event-driven architecture patterns including pub/sub, event sourcing, and the saga pattern — with Mermaid sequence diagrams.

Event-Driven Architecture: Patterns & Pitfalls

Event-Driven Architecture (EDA) is a software design pattern where the flow of the program is determined by events — significant changes in state that are published, detected, and consumed asynchronously.

Core Patterns

1. Pub/Sub (Publish–Subscribe)

Producers publish events to a topic; consumers subscribe independently. Producers don't know about consumers.

Rendering diagram…

Key property: Adding a new subscriber (e.g., Fraud Detection) requires zero changes to the publisher.

2. Event Sourcing

Instead of storing current state, store every state-changing event. Current state is derived by replaying events.

Traditional CRUDEvent Sourcing
UPDATE orders SET status='shipped'Append OrderShipped{orderId, carrier, trackingNo}
Current state in a rowCurrent state = replay of all events
History lostFull audit trail
One modelSeparate write (events) and read (projections) models

3. Saga Pattern — Coordinating Distributed Transactions

When a business process spans multiple services, there's no distributed transaction. Instead, a saga coordinates via events and compensating actions.

Rendering diagram…

Common Pitfalls

  1. Event schema evolution — Consumers must handle old and new event versions. Use a schema registry.
  2. Ordering guarantees — Events may arrive out of order across partitions. Use partition keys carefully.
  3. Idempotency — At-least-once delivery means handlers must be idempotent (use eventId deduplication).
  4. Debugging complexity — Distributed traces are harder than a single call stack. Invest in correlation IDs and observability.
  5. Eventual consistency — Users may see stale reads. Design UX to accommodate this.

When to Use EDA

  • Microservices needing loose coupling.
  • Real-time data pipelines (Kafka, Pulsar).
  • CQRS — separate read and write models.
  • Audit-critical systems (finance, healthcare) where event sourcing provides a full history.

Further Reading