PGSpider
github.com
- Free plan
- Linux
- Self-hosted
What PGSpider is
PGSpider is a free, self-hosted SQL cluster engine for accessing distributed data through Foreign Data Wrappers. It can combine similarly structured tables from multiple sources into a Multi-Tenant virtual table that clients query with one SQL statement. Queries and result fetching run across child nodes in parallel, and the engine can push WHERE clauses and aggregation functions down to those nodes. Supported aggregates on Multi-Tenant tables include AVG, STDDEV, and VARIANCE. Such tables also support INSERT, UPDATE, and DELETE, plus bulk INSERT using a batch_size option. The README demonstrates PostgreSQL and SQLite and lists PostgreSQL, MySQL, Oracle, GridDB, PGSpider, InfluxDB v2.0, and Amazon S3 Parquet storage for migration. PGSpider follows PostgreSQL-like operation and documents use with libpq and psql. It runs on Linux; deployment instructions describe building from source and installing to /usr/local/pgspider. The included license allows use, copying, modification, and distribution without a fee, while disclaiming warranties and any obligation to provide support or updates.
Who it suits
PGSpider suits teams that want to query distributed data through Foreign Data Wrappers using SQL. Its self-hosted Linux deployment and source build suit operators able to manage installation themselves.
Good to know, and what to watch
- Combines similarly structured sources in one virtual table.
- Runs child-node queries and result fetching in parallel.
- Pushes WHERE clauses and aggregations to child nodes.
- Supports PostgreSQL-like clients including libpq and psql.
- Several modification patterns are unsupported.
- Modifications are recommended in auto-commit mode.
- The tree-structure feature is in beta and not tested enough.
- The license disclaims support and updates.
Plans
Every plan and pricePGSpider
Open source
The verdict
PGSpider can unify distributed sources for SQL queries and supports parallel execution and selected modifications. Review the stated modification limitations and transaction guidance before relying on it for write workloads.
What the maker publishes
- Client compatibility
- PGSpider uses PostgreSQL-like operation and documents use with client applications such as libpq and psql. source
- Cloud transfer
- Data Compression Transfer sends compressed data through a Cloud Function and requires a pgspider_fdw relay server with endpoint and relay options. source
- Data modification
- Multi-Tenant tables support INSERT, UPDATE, and DELETE, as well as bulk INSERT using a batch_size option. source
- Data sources
- The README demonstrates PostgreSQL and SQLite and lists PostgreSQL, MySQL, Oracle, GridDB, PGSpider, InfluxDB v2.0, and Amazon S3 Parquet storage for migration. source
- Deployment
- The README documents building from source and installing to /usr/local/pgspider. source
- License
- The included license permits use, copying, modification, and distribution without a fee, and disclaims warranties and an obligation to provide support or updates. source
- Modification limitation
- RETURNING, WITH CHECK OPTION, and ON CONFLICT are unsupported with modification, and foreign-partition COPY and modification are unsupported. source
- Parallel queries
- PGSpider runs queries and fetches results from child nodes in parallel. source
- Purpose
- PGSpider is a SQL cluster engine for distributed big data that accesses data sources through Foreign Data Wrappers. source
- Pushdown
- It pushes WHERE clauses and aggregation functions down to child nodes, and supports AVG, STDDEV, and VARIANCE on Multi-Tenant tables. source
- Transaction limitation
- The README recommends running modifications in auto-commit mode because modified data may not be readable within a transaction and non-auto-commit changes may be inconsistent. source
- Tree feature status
- The README says the tree-structure feature is not tested enough and is in beta. source
- Virtual tables
- It can present similarly structured tables from multiple data sources as one Multi-Tenant virtual table queried with a single SQL statement. source
Checked against the maker's pages on 4 Oct 2026.