Making tomorrow's code look like today's
summary
Claim: replace fragile overnight batch jobs with message-driven sagas that keep future-facing state in-process, using queuing systems and persisted saga state so discounts and reports update in real time. Demonstrates using Azure Service Bus, Amazon SQS (900 seconds SQS limit), RabbitMQ TTL queues and a binary-queue trick (28 queues ≈8.5 years) to schedule timeouts, shows deadlocks from DB transactions, explains retries plus outbox pattern and optimistic concurrency to make concurrent clicks deterministic, and reports that the approach scales and eliminates nightly runtimes.
AI-written summary · about
chapters (10)
- 0:00Future-proofing code
- 1:45Runtime behavior and time issues
- 10:14Race conditions in discounts
- 17:16Rethinking time with sagas
- 22:15How sagas work and benefits
- 30:44Timeouts in message queues
- 37:53Batch jobs to sagas
- 41:49IT as profit center
- 44:38Resources and closing
- 45:41Q&A
related talks
51mCode the Future, Now
01h 01mThe Unsung Hero of Modern Software: Asynchronous Messaging
36mRise of the Machines [B]
45mResilient Predictive Data Pipelines
01h 02mBuilding cloud-ready applications in .NET
59mBuilding Resilient Applications In Microsoft Azure
29mMaking tech boring to keep data exciting
18mScaling, what you need to know