Connect a Database
SAMSKARA connects to infrastructure you already run through profiles, managed from Settings:
- Database profiles — Postgres, MySQL, or other SQL sources. Once added, they appear in SQL Workbench’s schema browser and are queryable directly.
- Storage profiles — S3 or S3-compatible object storage, used as the backing store for ArrowLake tables.
- Runtime profiles — where notebook code actually executes (local Python process or an isolated Docker container per session).
- AnyStream profiles — Kafka, Redpanda, Pulsar, NATS, or Kinesis brokers, accessible via
sm.stream("name")for bounded, read-safe message fetching.
Adding a database profile
Section titled “Adding a database profile”- Go to Settings and open the connections/profiles section.
- Add a new database profile with your connection details. Credentials can reference an org secret (
sm:SECRET_NAME) instead of being stored in plaintext on the profile. - Save. The connection is now available in SQL Workbench and, if you scope it to a project, inside that project’s notebooks.
Using it in SQL Workbench
Section titled “Using it in SQL Workbench”Once connected, the schema browser lists the database alongside any ArrowLake catalogs — tables, views, and columns are browsable without writing a query first.
Using it in a notebook
Section titled “Using it in a notebook”Database credentials configured on a profile are made available to notebook code the same way secrets are — through sm.secrets or the connector helpers registered at session start, scoped to whichever project the profile is attached to.
Read-only guardrails
Section titled “Read-only guardrails”SQL Workbench enforces a read-only transaction policy and a configurable per-query timeout on every connector, so exploratory queries can’t accidentally write to a source system.