Consistency and Agreements in Distributed Systems
summary
Argues that two‑phase distributed transactions fail in real distributed environments and even add a 10x throughput drop in testing, then demonstrates a working e‑commerce example that replaces DTC with outbox/inbox + dispatcher, sagas/process managers and durable messaging via RabbitMQ and Cosmos. Shows code and a GitHub repo with a runnable demo, explains at‑least‑once delivery plus idempotent inbox checks, and walks failure modes (retries, cancels, compensating actions) and operational tradeoffs.
AI-written summary · about
chapters (10)
- 0:00Distributed systems complexity
- 1:14Single transactions to challenges
- 6:10Two-phase commit failures
- 13:31Outbox and inbox patterns
- 20:50Idempotency in distributed systems
- 23:16Microservices with outbox and dispatcher
- 33:06Service conversation design
- 39:50E-commerce order processing example
- 53:09Outbox pattern code walkthrough
- 59:40Key lessons for distributed systems
related talks
01h 02mAt Least Once
01h 00mLife Beyond Distributed Transactions: An Apostate's Implementation
01h 02mLife Beyond Distributed Transactions: An Apostate's Implementation
53mLife Beyond Distributed Transactions: An Apostate's Implementation
52mLife Beyond Distributed Transactions: An Apostate's Implementation
46mLost in transaction? Strategies to deal with (in-)consistency in distributed systems
01h 01mLost in transaction Strategies to manage consistency in distributed systems
53mEffective Microservice Communication and Conversation Patterns