Partition identifiers

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

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 2014
Month YYYY-MM 2014-01
Day YYYY-MM-DD 2014-01-17
Hour YYYY-MM-DD-HH 2014-01-17-13

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:

  • 2013-01-01|France
  • 2013-01-01|Italy
  • 2013-01-02|France
  • 2013-01-02|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:

  • 2014-01-25/2014-01-28 : Single DAY dimension, select 3 days

  • 2014-01-25-14/2014-01-28-15: Single HOUR dimension, select 73 hours

  • 2014-02/2014-03|FR/IT: One MONTH dimension and one discrete dimension, select a total of 4 partitions:

    • 2014-02|FR
    • 2014-02|IT
    • 2014-03|FR
    • 2014-03|IT