Overview
The Snowflake source synchronizes data from a Snowflake table or view into Reelevant. Use it to bring warehouse data into your personalization workflows.
Configuration
Required Fields
| Field | Description |
|---|
account | Your Snowflake account identifier (e.g. xy12345.eu-west-1). |
username | Snowflake login username. |
warehouse | The Snowflake warehouse to use for running queries. |
database | The database name. |
schema | The schema name. |
table | The table or view to synchronize. |
Authentication
Snowflake supports two authentication methods:
Authenticate using a username and password.| Field | Description |
|---|
password | Snowflake login password. |
Authenticate using a key pair (JWT-based authentication). This is the recommended method for production environments.| Field | Description |
|---|
privateKey | The RSA private key in PEM format. |
SNOWFLAKE_JWT) authentication, and the password field is ignored.To generate a key pair for Snowflake:openssl genrsa 2048 | openssl pkcs8 -topk8 -inform PEM -out rsa_key.p8 -nocrypt
openssl rsa -in rsa_key.p8 -pubout -out rsa_key.pub
ALTER USER my_user SET RSA_PUBLIC_KEY='<public_key_content>';
Optional Fields
| Field | Description |
|---|
query | A custom SQL query to filter or transform the data before import. When provided, the query is executed instead of reading the full table. |
Custom SQL Queries
When you provide a custom SQL query, Reelevant executes it against Snowflake instead of reading the entire table. This is useful for:
- Filtering rows (e.g.
SELECT * FROM my_table WHERE is_active = TRUE)
- Selecting specific columns
- Joining multiple tables
- Applying transformations
The query must be valid Snowflake SQL. Make sure the user has the appropriate role and permissions to access the referenced tables.
Incremental Sync
Snowflake supports incremental synchronization when used with the Keep Previous Data update mode.
| Field | Description |
|---|
timestampColumn | The column name used to compute the delta since the last fetch. |
WHERE clause to your query (or generates one automatically) to only fetch rows where the timestamp column is greater than the last recorded value.
The timestamp column is optional even in Keep Previous Data mode — if your Snowflake view already handles delta logic internally, you can omit it.
Data Type Handling
Snowflake data types are streamed with the following conversions:
| Snowflake Type | Behavior |
|---|
VARCHAR, NUMBER, FLOAT, BOOLEAN | Passed through as-is. |
DATE, TIMESTAMP_*, TIME | Fetched as strings (using fetchAsString). |
VARIANT, OBJECT, ARRAY | Fetched as JSON strings. |
How It Works
- Reelevant connects to Snowflake using the provided credentials and warehouse.
- If a custom query is provided, it is executed. Otherwise,
SELECT * FROM <table> is run.
- Results are streamed row-by-row for efficient memory usage.
- Fields are extracted and made available for mapping.
- On subsequent syncs, the data is re-read (or incrementally fetched if a timestamp column is configured).
Ensure the Snowflake user has the appropriate role to access the specified warehouse, database, schema, and table.
