How it works

Reltio Zero Copy integration with Snowflake enables Reltio Intelligent 360^TM to retrieve interaction data from external cloud data warehouses using a secure, zero-copy model. Rather than duplicating data, Reltio sends real-time queries to Snowflake and joins the results with existing profiles in the platform. This is made possible through a lightweight, connector-based architecture and a federated query engine that performs just-in-time lookups.

Authentication and connection

Reltio connects to Snowflake using public-private key authentication. Administrators configure the connection by providing account and schema details through the Reltio API. The platform stores Snowflake credentials securely using AWS Secrets Manager. The connection setup includes:

• Snowflake account ID and region
• Warehouse, role, and schema name
• Private key-based login for secure access

Federated query execution

When interaction data is required, for example, when a profile is loaded, Reltio dynamically constructs a query using filters such as customer ID. The federated query engine then sends this query to the connected Snowflake warehouse. Only the filtered result set is retrieved and passed to the platform for runtime use. No full-table scans or bulk data loads are performed.

Computation of joins and aggregations is pushed down to the Snowflake engine, ensuring that data processing happens at the source.

Mapping and data linking

Each Snowflake table must be mapped to a defined interaction type in Reltio using the API. During this mapping process, users specify:

• The Snowflake table name
• The column to use for ordering (e.g., TRANSACTIONTIME)
• Which column to use for linking to Reltio profiles (e.g., customer ID)
• Which fields to display and their corresponding labels

The join between external interactions and Reltio entities is performed using crosswalks. No entity data is extracted from Snowflake, only interaction records are queried.

Governance and security

This architecture ensures that data governance policies are maintained. Sensitive data remains in Snowflake and is not persisted in Reltio. All data transfers occur over secure, authenticated connections. Snowflake user roles and permissions are respected, allowing organizations to control exactly what data Reltio can access.

Performance considerations

Because queries are filtered and pushed down to Snowflake, the system retrieves only the necessary data for each use case. No full-table replication or ETL is required. This model is optimized for:

• Low-latency profile enrichment
• Reduced storage and infrastructure cost
• On-demand access to large datasets without movement