Frequently Asked Questions

How is the model metadata generated?

Model statistics are generated using the Smart Insights metadata connection that connects the on-premise SAP HANA system to SAP Analytics Cloud, typically through the Cloud Connector.

The New HANA Live Connection dialog contains an Advanced Features to manage the metadata connection.

When model metadata is triggered manually from the Generate Model Metadata dialog, the statistics are generated one model at a time. This helps to balance the workload on the live SAP HANA system. The statistics are calculated using InA queries. The queries are executed using SAP HANA's sophisticated MDS component (Multi-Dimensional Services), similar to the way most SAP Analytics Cloud BI queries are executed. It typically takes between a few seconds and a few minutes to generate the model metadata for one model, depending on the size of the model, both in terms of data and number of dimensions.

What triggers the deletion of model metadata?

Deletion of cardinality statistics is triggered by the following actions:

Deletion of a model.
Deletion of a connection also deletes metadata for all models in that connection.
When a user with the maintain privilege on the live connection changes the Enable model metadata generation checkbox from checked to unchecked.
When the data source of a model is changed.

What exactly is stored?

The model metadata stored is the minimum required to calculate the most interesting candidate dimensions or hierarchical dimension members for Smart Insights. For a model, SAP Analytics Cloud stores the following for each dimension:

Dimension ID - to refer to a particular dimension.
Cardinality - the actual statistic needed for the Smart Insights algorithm. It represents a count of the number of members in the dimension.
Dimension type - flat, hierarchy, or attribute dimension.
Member ID - stored for suitable candidates in hierarchical dimensions.

What happens if my data has changed?

Data changes can affect the model metadata, so we recommend that you manually generate model metadata periodically. We also recommend that you manually generate model metadata if, for example, the underlying SAP HANA modeling view has changed structure or has new tables.