Type of data source in which hierarchies and aggregations have already been created by the cubes designer in advance.
They’re powerful as they can return information often quicker than relational data sources. The reason for this is that as it is pre built the definitions remain static and are only needed to be called when needed as opposed to built and then called. There are several data sources supported in tableau;
1. Oracle Essbase
2. Teradata OLAP
3. Microsoft Analysis Services
4. SAP Netweaver business Warehouse
5. Microsoft PowerPivot
When working with a cube data source the new calculated members can be created using MDX formulas instead of the regular calculated fields in tableau. In this new format they can either be a new calculated measure or a calculated dimension member (a new field within an existing hierarchy) so that if you have a Product with three members (Heels, Sneakers and Hats) a new calculated member defined as Shoes can be the sum of the Heels and Sneakers. Once this is created when showcasing the Products dimension on the view, the rows displayed are Heels, Sneakers, Hats and Shoes. This new member can be used the same as any other field in the view.
It is worth noting that when using a cube data source, some tableau features don’t work in the same manner as through a regular data source. Below are some of the features which don’t work in the same way.
Actions – Drill down actions defined in the cube are not available in tableau. Cube data also does not allow actions from other cube data sources, in the case where it is needed for cubes to interact with other cubes the cube structure must be altered to include the other data needed.
Advanced Analytic Features – LOD’s, trend lines, forecasting and clustering are not supported in cube data sources.
Aggregate Calculation Functions - Since the data is pre-aggregated it does not allow any further relational aggregation (SUM(), AVG() etc.) inside the calculated fields. If needed however, cube data sources support table calculations which allow more flexibility within the view.
Aliases – these can only be created by the cube’s designer.
Bins – It is not available for measures the same way it is for relational data sources, it is possible to create a calculated field which replicates same bin structure.
str((INT([Measure]/1000)) * 1000)
Data Blending – Cube data sources must be the primary data source for blending data.
Date Dimensions – in these use cases they are mainly organised into hierarchies which contain different level.
Extracts – Cannot be created from most cube data sources as cube and relational data sources have incompatible data structures. When creating extracts these are stored in relational data sources, which has a different structure that does not support a cube source.
Filters – Some filter card options are not available; cube attributes can be used as filters to show a singular level within the hierarchy but they cannot be used to slice the data.
Groups – as previously mentioned, the data must be already pre-aggregated in the cube data source and no further editing can be done in tableau. Multidimentional expressions can be made using a calculated member to create a group.
Parameters – Parameter values cannot be used to filter dimensions in an MDX calculation.
Publishing – This data source can be published to Tableau Server but doesn’t support pass-through connections meaning that to use the data source it must be downloaded locally and used in tableau desktop. This format is not supported in Tableau Cloud.