Database connections

DataMaker can push generated rows directly into a database, no INSERT script in the middle. We support six engines today.

Engine	Driver / protocol	Notes
PostgreSQL	libpq, TLS	12 — 17. RDS / Aurora / Supabase / Neon work.
MySQL	mysql-connector, TLS	8.x. PlanetScale, Aurora MySQL work.
MongoDB	MongoDB wire protocol, TLS	4.4+. Atlas works (use the SRV URL).
MSSQL	TDS, TLS	2017+. Azure SQL works.
Oracle	OCI thin / thick	19c+. Wallet auth supported.
IBM DB2	DRDA	LUW 11.5+. z/OS via DRDA.

Create a database connection

Project → Connections → New → Database, then pick the engine. Required fields:

Name — what you’ll see in the UI and in API responses.
Host and port.
Database name (or schema for some engines).
Username + password, or a managed secret reference.

Optional:

TLS — force TLS, accept self-signed, attach a CA.
SSH tunnel — for engines that aren’t public-internet-reachable.
Schema (Postgres / Oracle / DB2) — default schema for unqualified table names.

Click Verify. We open a TCP connection, authenticate, and run SELECT 1 (or the equivalent for MongoDB). Verified connections show a green dot in the sidebar.

Push a template

Open a template, click Generate → Push to….
Pick the connection. We discover the available tables / collections.
Pick a target table. The mapping screen suggests column matches by name; override any that are wrong.
Pick a mode:
- insert — straight INSERT (or document insert).
- upsert — INSERT ... ON CONFLICT (key) DO UPDATE (Postgres / MySQL), MERGE (MSSQL / Oracle), replace (MongoDB).
- truncate-then-insert — replace all rows. Use with care.
Generate. Per-row success/failure is reported in the run log.

Push from a scenario

The same target is reachable from a Python scenario via the SDK:

from datamaker import DataMaker

dm = DataMaker()                                      # picks up DM_API_KEY
customers = dm.generate(template_id="tmpl_customer", count=500)

dm.connection("conn_postgres_dev").insert(
    table="customers",
    rows=customers,
    on_conflict="update",
    key="id",
)

For more, see Scenarios → Scenario API and API & SDKs → Python SDK.

Performance & batching

DataMaker batches inserts (default: 500 rows per batch). For very large pushes (>100k rows), prefer:

Postgres: COPY mode. Roughly 5× faster than batched INSERT.
MongoDB: bulk_write (default).
MSSQL: BULK INSERT with a temporary table swap.

Set the mode in the push dialog or via mode="copy" / mode="bulk" in the SDK.

Rolling back a push

DataMaker tracks every push as a run, with the IDs of inserted rows. From the run detail page you can rollback — DataMaker deletes the rows it inserted (matching by the key you pushed under). Rollback is best-effort; if downstream triggers / cascades have modified the rows, the delete may need a manual override.

Auth providers

For managed cloud databases, prefer IAM-style auth where possible:

AWS RDS / Aurora: IAM authentication tokens.
Azure SQL / Postgres: Microsoft Entra ID (formerly AAD).
Google Cloud SQL: IAM database authentication.

In the connection form, pick Auth → Cloud IAM and follow the engine-specific setup guide.