How do I load data from Snowflake?

You want to bring tables from a Snowflake database into Keboola so you can transform, combine, and use that data. In Keboola this is done with the Snowflake data source connector (a query-based database connector). This page takes you from nothing to data landing in Keboola Storage.

Time: ~15 minutes · You will need: a Keboola project and a Snowflake account with rights to grant access.

Before you start

Make sure you have:

• A Keboola project you can sign in to, with rights to create configurations.
• Snowflake connection details: account identifier (host), warehouse, database, and schema you want to read from.
• Permission in Snowflake to run GRANT statements (or a colleague who can). You will create a dedicated read-only user for Keboola in Step 1.

If your Snowflake instance is not reachable from the public internet, you will also need to set up an SSH tunnel. That is covered separately; the steps below assume a directly reachable host.

Step 1 — Create a read-only Snowflake user for Keboola

Always give Keboola its own user with read-only access, rather than reusing a personal login. Run the following in Snowflake, replacing the MY_* placeholders with your warehouse, database, and schema names and choosing a password:

CREATE ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR;
GRANT USAGE ON WAREHOUSE MY_WAREHOUSE TO ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR;
GRANT USAGE ON DATABASE MY_DATA TO ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR;
GRANT USAGE ON SCHEMA MY_DATA.MY_SCHEMA TO ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR;
GRANT SELECT ON ALL TABLES IN SCHEMA MY_DATA.MY_SCHEMA TO ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR;
GRANT SELECT ON ALL VIEWS IN SCHEMA MY_DATA.MY_SCHEMA TO ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR;
CREATE USER KEBOOLA_SNOWFLAKE_EXTRACTOR PASSWORD = 'MY_PASSWORD' DEFAULT_ROLE = KEBOOLA_SNOWFLAKE_EXTRACTOR MUST_CHANGE_PASSWORD = FALSE TYPE = LEGACY_SERVICE;
GRANT ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR TO USER KEBOOLA_SNOWFLAKE_EXTRACTOR;

This grants access to the tables and views that exist now. Tables created later will not be readable until you re-run the two GRANT SELECT ON ALL ... statements (or grant the role broader future privileges).

Step 2 — Add the Snowflake connector in Keboola

1. In your Keboola project, open Components and choose Add Component (Components → Data Source Connectors).
2. Search for Snowflake and select the Snowflake data source connector.
3. Choose Add Configuration, give it a name that describes the source (for example, Snowflake — Sales DB), and confirm.

You now have an empty configuration, ready for credentials.

Step 3 — Connect to your Snowflake database

1. In the new configuration, click Set Up Credentials First.
2. Enter the connection details for the KEBOOLA_SNOWFLAKE_EXTRACTOR user from Step 1: host/account, warehouse, database, schema, username, and password.
3. Click Test Credentials, then Save.

If the test fails, see Troubleshooting below before continuing.

Step 4 — Choose the tables to load

After the credentials are saved, the connector automatically lists every table and view the user can read.

1. Select the tables you want to bring into Keboola.
2. Click Create.

Each selected table becomes a row in the configuration. You can:

1. Add another table later with New Table.
2. Open any table to edit what it loads, or disable it so it is skipped when the whole configuration runs.
3. Change the connection later via the Database Credentials link.

Step 5 — Run it and confirm the data landed

1. Click Run on the configuration.
2. When the job finishes successfully, open Storage and find the bucket created for this connector.
3. Confirm your tables are there with the expected row counts.

You have now loaded data from Snowflake into Keboola.

Load only new rows next time (incremental fetching)

By default each run reloads the full table. For large tables that only grow or get updated, switch on incremental fetching so each run reads only changed rows:

1. Open the table in the configuration
2. Set the ordering column — either an always-increasing ID or a last-modified timestamp.
3. Save and run

Keboola records the highest value it has seen in the configuration state and, on the next run, fetches only rows beyond that value. Incremental fetching suits tables where rows are added (and optionally updated) but not hard-deleted; it cannot detect DELETE statements. If you need to capture deletes or every change, review the Change Data Capture options.

Schedule it to run automatically

To keep Keboola in sync, add this connector to a Flow and set a schedule (for example, hourly or nightly). The connector runs in scheduled intervals (micro-batching) rather than streaming changes in real time.

Troubleshooting

Symptom	Likely cause	Fix
Test Credentials fails to connect	Wrong account/host, or host not reachable	Re-check the account identifier; if the database is private, set up an SSH tunnel.
Connects, but no tables are listed	The role lacks USAGE/SELECT on the schema	Re-run the GRANT statements from Step 1 for the correct database and schema.
A new table is missing from the list	Grants only covered tables that existed at grant time	Re-run the two GRANT SELECT ON ALL ... statements, then refresh.
”insufficient privileges” during a run	Warehouse usage not granted	Confirm GRANT USAGE ON WAREHOUSE ... TO ROLE KEBOOLA_SNOWFLAKE_EXTRACTOR.

Related

• Database data source connectors — overview, SSH tunnel, and CDC options.
• SQL database connectors — configuration reference — advanced mode and server-specific notes.
• Tutorial: Loading data from a database — guided walkthrough.