Free to read ยท Three levels
RabbitMQ: routing, responsibility and dependable message processing
Eighteen developed lessons connect AMQP 0-9-1 topology and message ownership to quorum queues, bounded retries, Streams and observable recovery.
Start learningIn this course
Learn through synthetic parcel jobs, reply correlation, retained event fixtures and bounded resource calculations. Six lessons at each level develop specific design decisions with explained examples, failure diagnosis and worked exercises.
The course separates publisher acceptance, routing, consumer delivery, external business effects and acknowledgments. It makes identity and responsibility explicit so retry, failover and replay can be reviewed without promising guarantees outside the actual system boundary.
What you will learn
- Design exchanges, queue identities, binding scope and bounded consumer responsibility.
- Reason through confirms, acknowledgments, quorum availability, retries and external-effect duplication.
- Review Streams, access, resource policies, topology transitions and recovery using explicit version and evidence boundaries.
Your learning path
Three levels
Basic: protocol resources, routing and ownership
Build precise message, topology, confirm, acknowledgment and lifetime contracts.
- Messaging, connections and channels
- Exchanges, queues, bindings and routing
- Publisher confirms and unroutable messages
- Consumer acknowledgments and redelivery
- Prefetch, work and ordering
- Durability, queue types and message properties
Intermediate: replicated work, retry safety and operational evidence
Connect quorum health, bounded retries, outbox handoffs and access to measurable outcomes.
- Quorum replication and write availability
- TTL, dead-lettering and bounded retries
- Duplicate-safe workers and the outbox
- Heartbeats, connections and recovery
- Virtual hosts, permissions and TLS
- Monitoring ready, unacked and backpressure
Advanced: ordered ownership, retained replay and accepted recovery
Review sequencing, RPC, Streams, policies and version transitions with bounded evidence.
- Single active consumer and key ordering
- RPC correlation, timeouts and replies
- Streams, retention and offset replay
- Policies, limits and resource budgets
- Topology migration and upgrades
- Failure schedules and recovery acceptance
Who this is for
- Application developers designing routed jobs, subscriptions and duplicate-safe workers.
- Engineers reviewing message loss reports, stalled consumers, retry loops and recovery evidence.
Before you start
- Basic programming, JSON and familiarity with application calls, durable storage and asynchronous work.
- Python standard-library examples are isolated models; no RabbitMQ installation, client package, database or cloud account is required to study them.