Event-Driven Architecture

Choosing Event-Driven vs Request-Driven

A practitioner's decision framework for knowing when event-driven architecture pays off—and when it just adds complexity.

Event Modeling

How to name, size, and classify events so your system communicates facts clearly instead of leaking implementation details.

Topology Patterns

A clear-eyed comparison of pub/sub, event bus, broker, and mesh topologies—and which to reach for in what context.

Schema and Contracts

How to define, register, and evolve event schemas using Avro and Protobuf so producers and consumers never break each other.

Outbox and Inbox

How the transactional outbox and inbox patterns ensure events are reliably published and processed exactly once, without two-phase commit.

Replay and Reprocessing

When and how to replay events—covering offset management, projection rebuilds, and the operational runbook for safe reprocessing.

Idempotency, End to End

How to design deduplication keys, idempotency stores, and retention policies so that processing the same event twice never corrupts your system.

Multi-Tenant Event Flows

How to partition, isolate, and protect multi-tenant event streams so that noisy neighbors and data leakage are design-time non-issues.

Observability for Streams

How to instrument event-driven systems for consumer lag, throughput, dead-letter queues, and end-to-end distributed tracing.

Anti-Patterns

The most damaging ways teams misuse event-driven architecture—distributed monoliths via events, event-as-RPC, and schema sprawl.