Snowflake data types

In Stream Pipelines, tables are not created in the database automatically. When you create tables in Snowflake, the column names and data types must match the Data Fabric object properties of the corresponding objects that are used in Stream Pipelines.

This table shows the object properties data types, as defined in Data Catalog, and their corresponding Snowflake data types:

Category Data Catalog data type Snowflake data type Snowflake comment
Date and Time data types Datetime:
  • Standard
  • American
  • Three-character month
 DATETIME Alias for one of the TIMESTAMP variations. By default, TIMESTAMP_NTZ.

TIMESTAMP_LTZ with the local time zone. The time zone, if provided, is not stored.

TIMESTAMP_NTZ with no time zone. The time zone, if provided, is not stored.

TIMESTAMP_TZ TIMESTAMP with the time zone.
Datetime:
  • Other
 STRING
Date:
  • Standard
  • American
  • Basic
 DATE
Time:
  • Other
 STRING
Time:
  • Standard
 TIME (by default, NTZ)
Number NUMBER Default precision and scale are (38,0). You can change precision.
Numeric data types Datetime:
  • Epoch
 NUMBER
Integer:
  • Maximum: Less than 32767
  • Exclusive maximum: Less than 32768
  • Minimum: More than -32767
  • Exclusive minimum: More than -32768
  • Multiple of (number of digits)
 NUMBER Synonymous with NUMBER, but precision and scale cannot be specified.
Integer:
  • Maximum: Between 32767 and 2147483647
  • Exclusive maximum: Between 32768 and 2147483647
  • Minimum: Between -2147483647 and -32767
  • Exclusive minimum: Between -2147483648 and -32768
  • Multiple of (number of digits)
  • None of the above is provided
 NUMBER
Integer:
  • Maximum: More than 2147483647
  • Exclusive maximum: More than 2147483648
  • Minimum: Less than -2147483647
  • Exclusive minimum: Less than -2147483648
  • Multiple of (number of digits)
 NUMBER
Number NUMBER Default precision and scale are (38,0).
Logical data types Boolean BOOLEAN
String STRING Synonymous with VARCHAR.
Object STRING
Note: Data Catalog object data types are supported for use with objects that have localized strings that are defined as part of the Data Catalog Locale Selections feature. Currently, Stream Pipelines processes only the first value of the object values. For example, if the localized value within the NDJSON record is {"desc": {"en_US":"Blue Car", "de_DE": "Blaues Auto", "es_ES": "Coche Azul"},, then only the first value, "Blue Car", is stored in the table column.