Sharding Without the Complexity

on YouTube ↗31m

" summary

Citus is a PostgreSQL extension that shards tables using a coordinator-and-workers architecture with a default of 64 shards per table. It installs at the system level and is enabled per database; reference tables are replicated to all workers, colocalized tables and a push-down planner optimize distributed joins and aggregates, COPY runs in parallel across shards, logical replication enables live rebalancing with zero downtime, and workers can be added or removed online.

AI-written summary · about

" chapters (9)

1. 0:00Introduction
2. 2:30Relational databases and sharding
3. 5:18Multi-entity approaches
4. 8:11Partitioning and performance
5. 13:51Citus Data
6. 17:39Citus technical overview
7. 21:11Citus benefits and use cases
8. 24:52Availability and disaster recovery
9. 27:07Q&A

" related talks

• 44mDistributed Postgres with CitusWill Leinweber · HighLoad++ 2017
• 08mFoundations of Distributed PostgreSQL with CitusOnder Kalaci · Azure Cosmos DB: Liftoff 2022
• 31mKEYNOTE: The Distributed PostgreSQL Problem & How Citus Solves itMarco Slot · Citus Con: An Event for Postgres 2023
• 28mEfficiently distributing Postgres with Citus – How to choose the right shard key?Sai Krishna Srirampur · Citus Con: An Event for Postgres 2022
• 23mFollowing Select Statements Through Citus InternalsOnder Kalaci · Citus Con: An Event for Postgres 2022
• 33mCitus & Patroni: The Key to Scalable and Fault-Tolerant PostgreSQLAlexander Kukushkin · Citus Con: An Event for Postgres 2023
• 28mCitus on KubernetesÁlvaro Hernández · Citus Con: An Event for Postgres 2023
• 16mKeynote – Umur CubukcuUmur Cubukcu · PGConf Silicon Valley 2015