Amazon S3 Strong Consistency
summary
Amazon S3 now guarantees strong read‑after‑write consistency for all operations with no change in price, performance, or regional isolation, across exabytes of data, ~100 trillion objects, and tens of millions of requests per second. The engineering team proves correctness using formal methods: protocol‑level deductive proofs, detailed P state-machine models analyzed by model checking, and code conformance checking that runs an executable reference model alongside implementation; billions of states explored daily on AWS compute, and the feature is deployed and operational.
AI-written summary · about
chapters (5)
- 0:00Introduction to S3 strong consistency
- 4:43Why strong consistency is hard
- 6:09Automated reasoning for correctness
- 12:51Model checking at S3 scale
- 21:23Conclusion and Q&A
related talks
52mHow We Survive Split-Brain and Keep Writing Data via the S3 Protocol
35mBuilding Systems that Last
01h 15mCAP Theorem — two decades and few clouds later
01h 08mTransactions and Scalability in Cloud Databases—Can’t We Have Both?
52mEventual Consistency in a Stateful Service
14mBolt-On Strong Consistency: Specification, Implementation, and Verification
24mDistributed Transactions at Scale in Amazon DynamoDB
33mGoogle Cloud Spanner: Global consistency at scale