Federating Queries Using postgres_fdw
summary
Sets up postgres_fdw on PostgreSQL 9.3, creates foreign servers, user mappings and foreign tables, then federates with union all views over eight nodes in a round‑robin demo (counts across eight nodes finish in ~3 seconds on a laptop). Builds work but exposes limits: foreign tables have no constraint exclusion, executor issues sequential remote scans, joins and averages pull raw rows back, and network/trip overhead is real. Recommends careful node/table placement, tuning, and staging before committing.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 2:44Limitations
- 7:12Creating a federated view
- 8:47General federated strategies
- 13:00Demo
- 15:00Network overhead
- 17:30Dumb queries
- 19:15Grouping
- 20:12Bubble joins
- 22:00Keys
- 24:45Check constraints
- 30:13Large working set
- 32:19Nodelevel partitioning
- 35:19Distributed processing
- 37:04Distributed replication
- 38:28Advice
- 41:22Questions
- 46:34Use Cases
- 49:46Is it Worth It
related talks
52mSharding PostgreSQL using Foreign Data Wrappers
34mJoining Heterogeneous Databases is a reality, not a Myth
46mWriting a Foreign Data Wrapper
50mBeyond Foreign Data Wrappers: building a data warehouse from heterogeneous external data sources
39mPostgreSQL: Joining Heterogeneous Databases is a reality, not a Myth
34mPostgreSQL sharding, state of the art
27mMigrating from MSSQL to PostgreSQL
25mRuby and PostgreSQL, a love story
