Search code examples
google-bigquerytableau-apiolap

Can you create preaggregated Dimensions/Measuresments like OLAP in BigQuery with Tableau?


During the Cloud Migration of an On-Premise Microsoft SQL DB, the OLAP Cube, which is part of it, should also be replaced (but not migrated directly). There is the business requirement to keep the functionality in Tableau that you can select different measurements and dimension with their corresponding aggregations, as is possible now when connecting to the OLAP Cube in Tableau.

The underlying Data Source View includes ca. 10 tables (e.g. customer, sales, payment-method, customer-segmentation, time). So via OLAP the analysis "give me the average sales per payment method per customer-segment for every week" is a couple of clicks, in pure SQL it's already some effort.

How can you offer defined aggregations for some BigQUery tables without the user having to write the joins and aggregations by themselves, mainly because it takes much more time than simply drag & drop (SQL skills & time of query-execution are not the issue)?


Solution

  • The answer turns out to be pretty straight forward:

    Join all source data together and write it into one flat table in BigQuery which includes the same information as the data source view in the OLAP Cube. Then Tableau connects to this table. The "measurements" logic from the cube is implemented as calculations in Tableau, the table columns are the dimensions.

    Some caution needs to be applied when replicating the measurements because 1:n relations in the Data Source View result in multiplied data in the flat table. This can be solvedwith the correct use of Distinct Functions (e.g. "Distinct Count") in the measurement definition.

    The table will end up quite large, but the queries on it are very fast, resulting in a performance increase compared to the OLAP Cube with the same user experience as using a cube in Tableau.