Show TOC

Integration of Physical PartitioningLocate this document in the navigation structure

Concept

You can physically partition a semantically partitioned object if the DBMS being used allows this.

More information: Partitioning

To do this, you partition the reference structure. The physical partitions are derived during generation of the semantic partitions.

There are three different scenarios for derivation of the physical partitions:

  • The semantic partitioning criterion is not a time characteristic ( Region (0REGION) for example)

  • The semantic partitioning criterion is a time characteristic ( calendar year (0CALYEAR) for example)

  • The semantic partitioning criteria contain both a time characteristic and a non-time characteristic (0CALYEAR and 0REGION for example)

Characteristic as Semantic Partitioning Criterion

If the semantic partitioning criterion is a characteristic, the physical partitioning criteria for each partition is taken directly from the reference structure and is therefore the same for all partitions.

Example:

A semantically partitioned InfoCube has the following properties:

  • Semantic partitioning: Region (0REGION); Values: AMERICAS, EMEA, APJ

  • Physical partitioning: Calendar month (0CALMONTH); Values:

The partitions would look like this:

Time-Characteristic as Semantic Partitioning Criterion

There are two different scenarios if the semantic partitioning criterion is a time characteristic:

  • The values for the time criterion in the semantic partitioning are not consecutive.. In this case, physical partitioning is not supported.

  • The values for the time criterion in the semantic partitioning are consecutive..

    For semantically partitioned DataStore objects, this means that the value range of the physical partitioning is identical to the value range of the semantic partitioning.

    For semantically partitioned InfoCubes, the calculation is more complicated:

    For InfoCubes, you can also define a maximum number of partitions. The system then starts by calculating the partition density. With calendar month (0CALMONTH) as the partitioning criterion for example, this means that the system calculates how many months a (physical) partition could contain. This is then applied to the semantic partitions. If the maximum number is then exceeded for a semantic partition, the system adjusts this. The partition could then contains five months for example instead of three months like the other partitions.

Time Characteristic and Non-Time Characteristic as Partition Criterion

If the semantic partitioning is based on both a non-time characteristic and a time characteristic, each partition is analyzed separately for the physical partitioning. The relevant rule is then applied for each semantic partition.

Example:

A semantically partitioned InfoCube has the following properties:

  • Semantic partitioning: Region (0REGION); Values: AMERICAS, EMEA, APJ

  • Physical partitioning: Calendar Year (0CALYEAR); Values: 2007

The semantic partitions would look like this:

In this example, the AMERICAS and APJ partitions are partitioned according to the rules for characteristics as a semantic partitioning criterion. The EMEA partition is partitioned according to the rules for time characteristics as a semantic partitioning criterion.