Description:
This property specifies the dimension order of the actual semantic model file. When a semantic model is processed, dimensions are arranged in decreasing order of cardinality, with the highest cardinality arranged first. This property overrides the cardinality order. Changing the order can affect semantic model size, cuboid count, and query performance.

Default value and behavior:
By default, the dimension order is computed dynamically during a full semantic model process. The dimensions are arranged in decreasing order of cardinality.

Value:
Dimensions can be reordered using the properties dialog box on the Kyvos web portal to define the order. This property enables users to move a lower cardinality dimension into higher precedence in the semantic model structure. Dimensions not selected in this property will be arranged based on the decreasing order of their cardinality.

Scope:
Semantic Model: If the property is set on a semantic model, then the value will override the connection level value for that semantic model's job.

Comes into effect:
A full semantic model process is required for this to take effect.

Dependencies and related properties:

• Semantic Model: kyvos.build.dimensions.materialize

• kyvos.build.dimensions.hop

Recommendations:
This property has a significant impact on both the size of a semantic model as well as the query performance. The query performance improves for dimensions that are selected at the higher order. However, the cuboid size and semantic model process time increase resulting in  slower query performance when higher cardinality dimensions are placed at lower order.

General recommendations include:

• Dimensions should be arranged in descending order by cardinality (this is the default behavior).

• Frequently filtered dimensions should have higher precedence.

• Frequently referenced dimensions should have higher precedence.

Example:
When working with Olympics data, the Athlete dimension has the highest precedence due to its highest cardinality. However, since other dimensions are queried and filtered more frequently, such as Country or Event, these dimensions can be moved higher to improve query performance. But, this may result in a larger semantic model.