Following Select Statements Through Citus Internals
summary
Citus extends Postgres planner and executor to produce a distributed plan made of task is a SQL command units that run on regular Postgres shards tracked in metadata (example: a 32‑shard table yields 32 tasks). Aggregates like AVG are computed by summing shard sums and counts on the coordinator. The executor parallelizes work with a per‑node connection pool, caches one connection per node, and adaptive executor mimics TCP Slow Start using 10 ms ramps and doubling connections to balance latency and concurrency.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 1:32Query processing in Postgres
- 5:26Query processing in Citus
- 8:54Citus query planning
- 13:34Citus query execution
- 21:45Takeaways about Citus internals
related talks
44mDistributed Postgres with Citus
31mSharding Without the Complexity
08mFoundations of Distributed PostgreSQL with Citus
33mCitus & Patroni: The Key to Scalable and Fault-Tolerant PostgreSQL
31mKEYNOTE: The Distributed PostgreSQL Problem & How Citus Solves it
16mKeynote – Umur Cubukcu
28mCitus on Kubernetes
38mPostgreSQL as a distributed computing platform