Beyond Pub/Sub - Advanced Messaging Patterns
summary
Explains concrete fixes: Transactional Outbox + database-stored outgoing messages plus handler-side deduplication and idempotency yields exactly-once processing in practice; recommends NServiceBus patterns but also names Azure Service Bus and RabbitMQ for topics/queues, and uses dead-letter queues, bounded prefetch and retries (immediate, delayed, exponential) for failures. Shows workflow coordination via sagas (orchestrator vs choreography), parallel latency reduction with scatter-gather (DHL/FedEx/UPS example), and elastic scaling with competing consumers plus backpressure and error-queue monitoring.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:18Why teams use Pub/Sub
- 6:15Pub/Sub limitations
- 10:53Why Pub/Sub isn't a system design solution
- 11:36Transactional Outbox for dual writes
- 16:32Idempotency for exactly once
- 21:11Beyond Pub/Sub workflows
- 27:13Sagas for long processes
- 31:47Orchestration vs choreography
- 37:42Choosing orchestration or choreography
- 38:24Sagas for out-of-order
- 40:29Scatter Gather parallel
- 45:06Scatter Gather vs Sagas
- 46:29Competing Consumers scaling
- 51:50Pub/Sub with Consumers
- 52:33Back pressure overload
- 54:38Retry strategies
- 58:32Error queue monitoring
- 59:33When to apply patterns
related talks
01h 03mBeyond Pub/Sub - Advanced Messaging Patterns
01h 02mMessaging Patterns for Modern Software Solutions
38mMessaging Patterns to Transform Your Cloud Architecture
53mEffective Microservice Communication and Conversation Patterns
55mImplementing scalable messaging patterns in Azure
56mDistributed Messaging patterns
27mApplication Integration Patterns for Microservices
52mLife Beyond Distributed Transactions: An Apostate's Implementation