Partition identifiers

When dealing with partitioned datasets, you need to identify or refer to partitions. Within a dataset, a partition identifier uniquely identifies a partition.

The identifier of a partition is made by concatenating the dimension values, separated by | (pipe)

Time dimension identifiers

The format of a time dimension identifier depends on the time dimension granularity

Period

Format

Example

Year

YYYY

2020

Month

YYYY-MM

2020-01

Day

YYYY-MM-DD

2020-01-17

Hour

YYYY-MM-DD-HH

2020-01-17-13

Some special keywords are also available:

  • CURRENT_HOUR

  • CURRENT_DAY

  • CURRENT_MONTH

  • CURRENT_YEAR

  • PREVIOUS_HOUR

  • PREVIOUS_DAY

  • PREVIOUS_MONTH

  • PREVIOUS_YEAR

Discrete dimension identifiers

The identifier of a discrete dimension value is the value itself. As such, discrete dimensions should only contain letters and numbers.

Multiple dimensions

For example, if you have a time dimension named “date” with a “DAY” granularity and a discrete dimension named “country”, then you could have partitions like:

  • 2020-01-01|France

  • 2020-01-01|Italy

  • 2020-01-02|France

  • 2020-01-02|Italy

Ranges specifications

In various locations in DSS, you can use a “partition range specification” syntax to refer to a set of partitions or values of a dimension.

For example:

  • In “Exact values” dependency function

  • When building datasets

  • When creating scheduled jobs

The generic syntax is:

  • PARTITION_SPEC = DIMENSION_SPEC|DIMENSION_SPEC|....

  • DIMENSION_SPEC =

    • DATE # Single Date (Time)

    • DATE/DATE # Date Range (Time)

    • Any # Actual Value (ExactValue)

    • Any/Any/Any/... # Several values (ExactValue)

Examples:

  • 2020-01-25/2020-01-28: Single DAY dimension, select 3 days

  • 2020-01-25-14/2020-01-28-15: Single HOUR dimension, select 73 hours

  • 2020-02/2020-03|FR/IT: One MONTH dimension and one discrete dimension, select a total of 4 partitions:

    • 2020-02|FR

    • 2020-02|IT

    • 2020-03|FR

    • 2020-03|IT