Create a filtered extract from a live connection on Tableau Server

by Laura Scavino

This week I have learnt how to customise a data extract from a live connection on Tableau Server. I also learnt how to blend the two data sources from Tableau Server, which had no common fields with a simple trick. This blog post is all about improving performance of the workbook but it also gives a quick trick to data blending.

The first step is to Extract the data, which comes from a server live connection, therefore might be really slow to work on. It is possible, as a proof of concept, to just create a duplicate of the Tableau Server data source and do some testing on it.

The squared icon shows the Tableau Server live connection, whereas the two cylindrical icons shows the extract data, one blue for the primary data source and one red for the secondary data source.

 

It is possible to create an extract with filters, in order to reduce its size and improve performance. In this case I have decided to create a Set with only 5 members and use it as a extract data filter.

I have also find out that it is possible to replicate the Set in a different data source, without looking among all the ID numbers one by one. Simply copy the values from the set in the primary data source and paste them in the secondary data source by creating another set. It is possible to paste the values in the custom value list where it is written: enter text to search or add, instead of writing down your ID from a data source to another one.

Therefore you create the custom extract in both data sources, which it doesn’t have to be the same. You can in fact add different filters in both of them to have the desired outcome. Once the extracts are ready, it is possible to blend them in a quicker way, given the fact that we are now working on a filtered extract. The two datasources however do not have much in common, therefore I decided to create a calculated field in each extract.

By looking at the Date field, I noticed they are looking at different aspects. I create a new date field in both data extracts and I insert only the Date field in the calculated field. The key thing is to call both calculations in both data extracts in the same way, even if they have different values inside. Just click on Data -> Edit Relationships and add the new created field as a relationship between the data sources. Remember also to select the same date type in the relationship, in this case I chose the Month Day Year format.

I hope you can find this helpful!