Beginner's Guide to Partitioning vs. Sharding in Postgres
summary
Explains when native Postgres partitioning versus sharding with the Citus extension actually speeds queries: partitioning gives partition pruning and data locality, parallel autovacuum across partitions, cheaper storage tiers and near-instant drop partition to expire data, but requires upfront partition key choice and maintenance (PGPartman, pg_cron). Sharding with Citus distributes shards across nodes, adds CPU/memory/disk and parallelism, offers shard rebalancing, object propagation, collocated tables and create_distributed_table, and can combine with partitioning for large time‑series.
AI-written summary · about
chapters (8)
- 0:00Introduction and speaker background
- 3:12What is Postgres partitioning
- 13:33What is sharding
- 25:34Partitioning vs sharding comparison
- 28:45Why partitioning improves query performance
- 32:19Why sharding improves query performance
- 35:34When sharding helps performance
- 36:15Key takeaways and conclusion
related talks
42mBeginner's Guide to Partitioning vs Sharding in Postgres
28mEfficiently distributing Postgres with Citus – How to choose the right shard key?
33mFive Sharding Data Models and Which is Right
44mDistributed Postgres with Citus
31mSharding Without the Complexity
31mPostgres Partitioning Best Practices
34mPostgreSQL sharding, state of the art
31mKEYNOTE: The Distributed PostgreSQL Problem & How Citus Solves it