A Shared nothing Cluster System Postgres XC

on YouTube ↗56m

" summary

Postgres-XC aims for write-scalability with a shared-nothing architecture using coordinators, data nodes and a GTM that hands out global transaction IDs and snapshots. It implements distributed vs replicated tables (hash/round-robin but no range or vertical partitioning), two‑phase commit and fast query shipping when replicas co‑locate, and uses a GTM proxy and pooler to reduce load. It delivers transparent Postgres clients and dbt1 performance gains but has a GTM bottleneck, no serializable global isolation, and no built‑in automatic failover.

AI-written summary · about

" chapters (12)

1. 0:00Introduction to Postgres-XC
2. 2:10Cluster architecture
3. 6:52Pros and cons of clustering
4. 10:24Data distribution methods
5. 16:08Query performance optimization
6. 20:40Parallelism and load balancing
7. 25:38Query shipping and explain plans
8. 31:25Optimizing join operations
9. 35:38Cost estimation limitations
10. 37:22Deadlock handling
11. 40:36ACID compliance and atomicity
12. 47:50Bottlenecks and future work

" related talks

• 47mWhat Is Postgres XC?Maksim Boguk · HighLoad++ 2012
• 50mSupersized Postgres: Postgres-XL for OLTP and Big Data AnalyticsMason Sharp · PostgresOpen 2014
• 58mPostgres-XC dynamic cluster managmentKoichi Suzuki · PostgresOpen 2013
• 49mPostgreSQL’s Journey to Distributed DatabasesSameer M, Steve A · Postgres Conference Orlando 2025
• 59mUsing Postgres XLRichard Silva · PGConf Silicon Valley 2016
• 41mRevisiting XTM: A Practical Case Study Highlighting Its NeedsYuya Watari · PGConf.dev 2025
• 51mEmbrace, Extend, Evolve - Making PostgreSQL Work for a Cloud Native WorldKarthik Ranganathan · Distributed SQL Summit Asia 2023
• 38mPostgreSQL as a distributed computing platformMarco Slot · PostgresOpen Silicon Valley 2017