A technically rigorous guide to event-driven architecture and the message broker landscape — with dry wit and honest assessments instead of vendor brochures.
Covers the fundamentals without pulling punches: core concepts, patterns (pub/sub, event sourcing, CQRS, sagas), schema evolution, error handling and delivery guarantees (including why "exactly-once" is a polite fiction), observability, security, testing, and a field guide to anti-patterns.
An opinionated evaluation of 16+ message brokers using a consistent framework — architecture, strengths, weaknesses, operational reality, code examples, and a verdict for each:
| Broker | One-Liner |
|---|---|
| Apache Kafka | The 800-pound gorilla |
| RabbitMQ | The reliable workhorse |
| Apache Pulsar | The multi-tenant challenger |
| AWS SNS/SQS & EventBridge | The AWS-native path |
| Google Pub/Sub & Azure Event Hubs | The other cloud-native options |
| Redis Streams | When your cache gets ambitious |
| NATS & JetStream | The lightweight speed demon |
| ActiveMQ & Artemis | The enterprise veterans |
| ZeroMQ | The brokerless broker |
| Redpanda | Kafka without the JVM tax |
| Memphis | The developer-experience play |
| Solace PubSub+ | The enterprise dark horse |
| Chronicle Queue | When microseconds matter |
| Aeron | For the ultra-low-latency crowd |
Plus a chapter on the obscure and curious (QStash, Watermill, EventStoreDB, RocketMQ, and more), a full comparison matrix, and a selection guide with decision trees.
Requires mdBook:
mdbook serve --openThanks to Georgiy Treyvus, CloudStreet Product Manager, for the idea behind this book.
See LICENSE.