Snowflake uses the Variant data type to store and represent semi-structured data. The Variant data type can store values of any other type, including object and array, up to a maximum size of 16 MB uncompressed.

Array, object, and variant from the Snowflake source are mapped to the String data type in Cloud Data Integration. While writing to the target, these strings can be written as Array, Object, or Variant columns to the Snowflake target. The strings that you write to Snowflake must be in a serialization format, just as they appear after the read operation.

When you use the Create New at Runtime option to write the Variant data type from a source to the Snowflake target, Data Integration writes Variant as Varchar to the target. You can edit the field mapping to map Varchar to Variant before you write to the target. In a completely parameterized mapping, you cannot edit the target metadata from the default Varchar data type to variant.

The following table lists the semi-structured data types that you can read from Snowflake and the corresponding transformation data types that these map to in Cloud Data Integration: