Easy Eventual Consistency with Actor Models + Amazon Web Services
summary
Claims a practical way to get strong eventual consistency in AWS by treating clickstream processing as immutable per-day sets, using hash IDs and set unions to merge replicas, and recovering from region outages by reconciling hashes. Uses CRDTs + Event Sourcing + immutability, actor systems (Akka.NET style) as competing consumers over SQS, backs data to S3 and Elasticsearch, runs ~10–20 EC2 instances, achieves near real‑time (sub-30s) metrics and successful multi-day recovery after an outage.
AI-written summary · about
chapters (29)
- 0:00Introduction
- 0:44Philips background
- 9:05Weak Eventual Consistency
- 9:38XP Analogy
- 10:23Simple Principle
- 10:45Conflictfree replicated data types
- 11:20Cap theorem
- 12:38CRDTs
- 13:37Querying
- 14:11Merge
- 15:17I dont understand
- 15:30Union is all you need
- 15:59Commutative
- 16:41Why Immutability
- 17:26How to Store
- 18:14What do you do
- 19:25ClickStream Pipeline
- 19:35ClickStream Events
- 20:03Multiple Actor Systems
- 20:32Actor Systems App
- 21:24Elastic Search
- 23:36Continuous Deployment
- 24:24Distributed Swarm
- 28:40Catastrophic Failure
- 29:10Data Recovery
- 30:08Last Batch Request
- 32:54Hash Sets
- 33:25SQS
- 34:51Questions
related talks
43mEffective Eventual Consistency with Actor Models + Amazon Web Services
39mManaging Consistency, State and Identity in Distributed Microservices
46mBuilding high available systems with Akka Distributed Data
51mImmutable data stores and CORS for safety, flexibility and profit
01h 11mEmbracing Messaging and Eventual Consistency in your Microservices Solutions
54mHow Events Are Reshaping Modern Systems
46mReliability Nirvana
26mValidation in an eventually consistent world