messaging-rabbitmq-architect_skill

Overview

This skill designs production-ready RabbitMQ architectures covering exchanges, quorum and stream queues, routing patterns, clustering, dead letter exchanges, and AMQP best practices. I produce clear topology recommendations, queue arguments, publisher/consumer settings, and error handling patterns. The outputs are actionable: topology diagrams, configuration snippets, and operational rules to implement immediately.

How this skill works

For each requested message flow I validate RabbitMQ version and functional requirements, select exchange types and queue types (classic/quorum/stream) and generate routing and DLX retry patterns. I recommend publisher confirms, delivery modes, consumer ack strategies and prefetch tuning. For clustering and high-throughput scenarios I provide quorum replication, stream queue configuration, and federation or consistent-hash sharding patterns.

When to use it

• Design a resilient production topology with replication and durability requirements
• Implement DLX-based retries and dead-letter handling for failed messages
• Choose between quorum queues and streams for durability vs throughput
• Plan a multi-node cluster with Raft quorum and appropriate node counts
• Shard high-volume workloads using consistent-hash exchanges or stream consumers

Best practices

• Default to topic exchanges for flexible routing unless exact one-to-one mapping is required
• Use quorum queues for production durability; streams for append-only, high-throughput use cases
• Enable publisher confirms and delivery_mode=2 for durability; avoid AMQP transactions
• Use manual consumer acks and start with prefetch=10, then tune based on processing profile
• Implement DLX + TTL retry queues and a delivery-limit policy to avoid infinite redelivery
• Run clusters with an odd number of nodes (3/5) and set x-quorum-initial-group-size accordingly

Example use cases

• Single-exchange topology: events topic exchange → quorum order-processing-queue with confirms and prefetch=10
• Multi-exchange routing: orders, payments, notifications exchanges with topic bindings and DLX for failed orders
• Retry/backoff: retry-order-5s and retry-order-30s TTL queues that route back to main queue, final DLX for manual inspection
• Clustering: 3-node cluster example, quorum queue replication and network/port requirements
• High-throughput streaming: stream queue config with x-max-age and consumer offset strategies for audit/event logs