SQL

What it is

A bucket’s tables are one transactional store that also answers analytical queries. You write ordinary SQL over the Postgres wire; the plane decides how to store and serve it. There is no engine to provision, no capability to switch on and no schema registry: create the bucket, connect, run CREATE TABLE.

   psql · psycopg · Prisma · SQLAlchemy · GORM · SQLx · node-postgres
                          │
                          ▼   pg.uk-lon-1.dodil.io:5432   dbname = <bucket>
                 ┌───────────────────┐
                 │  tables gateway   │   auth, TLS, per-bucket isolation
                 └─────────┬─────────┘
                           ▼
                 ┌───────────────────┐
                 │  your bucket      │   Delta Lake tables
                 │  s3://<bucket>/   │   (also readable by any Delta tool)
                 └───────────────────┘

It is a system of record. OLTP workloads such as a double-entry ledger run directly on it, with the same isolation contract you would build against Postgres: snapshot reads, SELECT … FOR UPDATE guards, serialization failures you retry. It is also a warehouse: GROUP BY, window functions, joins and CTEs run over the same rows with no copy and no sync.

Who it is for

• Application developers who want Postgres semantics without running Postgres. Your ORM’s migrations and CRUD work; the compatibility pages list the exact differences.
• Teams that ingest documents and want structured rows to appear automatically. A table can be bound to a pipeline template so that rows materialize as files land in the bucket — see Pipelines.
• Analysts and data engineers who want the rows in an open format. Every table is a real Delta Lake table in your bucket; external Delta readers can scan it.

Start here

Pick the question you have: