PostgreSQL’s Journey to Distributed Databases
summary
Shows how Spanner, Yugabyte, Cockroach, Cosmos DB for Postgres and Aurora actually implement distributed Postgres: Spanner uses 8 GB splits, TrueTime and Paxos for global commits; Yugabyte combines Aurora-like compute with Spanner-like storage using Raft and tablets with 3‑ or 5‑node replication; Aurora keeps six copies and commits with a 4-of-6 quorum in 10 GB protection groups; serverless v2 scales by 256 ACUs (512 GB) and Limitless shards writes; DSQL spins per-connection query processors and uses optimistic concurrency with synchronous cross‑region writes.
AI-written summary · about
chapters (12)
- 0:00Distributed SQL intro
- 2:07App architecture evolution
- 5:59CAP and PACELC theorems
- 8:29Distributed SQL requirements
- 12:00Google Cloud Spanner architecture
- 20:59YugabyteDB design and features
- 23:48CockroachDB architecture
- 24:31Azure Cosmos DB PostgreSQL
- 26:16Aurora Storage and serverless v2
- 31:13Aurora Limitless scaling
- 37:31Introducing Aurora DSQL
- 45:18Summary
related talks
01h 13mExtending PostgreSQL to Google Spanner architecture
42mPostgreSQL Distributed: Architectures & Best practices
24mNavigating Distributed Databases: Challenges & Opportunities
51mEmbrace, Extend, Evolve - Making PostgreSQL Work for a Cloud Native World
38mPostgreSQL as a distributed computing platform
26mNewSQL: concepts, realities and when you should embrace it
30mTechnical Comparison of YugabyteDB with Oracle and PostgreSQL
47mAnatomy of a Distributed SQL Database - YugabyteDB