PostgreSQL sharding, state of the art
summary
Postgres relies on postgres_fdw plus declarative partitioning for built‑in sharding, with concrete features added over versions: FDW since 9.3, aggregate pushdown and writable foreign partitions in v10, partition‑wise join and aggregate in v11, and bulk insert in v14. Examples show joins falling from 47s to 123ms, a GROUP BY from 49s to ~2s, a global count from 49s to 1.7s, and 1M row inserts from 466s to 5.5s with batching. Missing: partial aggregate pushdown, rebalancing tools, HA and distributed atomic commit.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:25Database Sharding
- 6:14PostgreSQL Built-In Sharding
- 14:33Improvements & Key Features
- 28:49Missing Pieces
- 33:44References
related talks
52mSharding PostgreSQL using Foreign Data Wrappers
45mThe Future of Postgres Sharding
47mpg_shardman: Sharding in PostgreSQL
45mThe Future of Postgres Sharding
51mUsing Postgres FDW with a sharded PostgreSQL Database
33mFive Sharding Data Models and Which is Right
42mBeginner's Guide to Partitioning vs Sharding in Postgres
53mBreakout Session