Supersized Postgres: Postgres-XL for OLTP and Big Data Analytics

on YouTube ↗50m

" summary

Postgres-XL claims a single scale-out Postgres that handles OLTP and MPP analytics, implements cluster-wide MVCC by handing out global transaction IDs via a GTM and reduces that bottleneck with a gtm proxy that parcels ID ranges. Data lives on data nodes, distributed by hash or replicated; DDL uses a distribute by clause and deployment uses pgxc_ctl for local clusters. Limits include no native automatic HA, blocking table locks during re-sharding, and no global distributed constraints; tests show near-linear scaling and a surprising 15% degradation for two-replica write setups, and a production StormDB cluster runs this code.

AI-written summary · about

" chapters (3)

1. 0:00Introduction to Postgres XL
2. 2:02Architecture and features
3. 47:23Community and roadmap

" related talks

• 54mScaling out by distributing and replicating data in Postgres XCAshutosh Bapat · Postgres Open 2012
• 59mUsing Postgres XLRichard Silva · PGConf Silicon Valley 2016
• 56mA Shared nothing Cluster System Postgres XCAmit Khandekar · Postgres Open 2012
• 47mWhat Is Postgres XC?Maksim Boguk · HighLoad++ 2012
• 39mPostgreSQL BI Performance & News from XL projectMark Wong · PostgresOpen Silicon Valley 2017
• 13mBuilding reliable infrastructure with PostgresAndrey Fleyta · PGConf India 2025
• 36mPostgres at any ScaleCraig Kerstiens · Birmingham on Rails 2020
• 48mHow to Tame A Mastodon: Lessons for PostgreSQL at ScaleDavid Christensen · PGConf NYC 2022