I have a client who has 2 databases, let's call them coke and pepsi. These two databases are on different servers and have different data connections.
Both databases have similarly named tables and field names.
For example, both the coke_db and the pepsi_db:
- have tables called attributes, country and sales
- the attributes tables has fields: product_id, name and size
- the country tables has fields: country and region
- the sales tables has fields: product_id, region and sales
The data connection joins these three tables. There are two data connections because one is for coke and one is for pepsi. The tableas and data connections have exactly the same design.
The data CANNOT be blended. At no time is it desirable to mix data from one data source and other. Rather, the desire is to create one view of the data that can load the data for either coke or pepsi.
I have tried doing this with Parameters but it is ATTR aggregating my dimensions. I've also tried it with Actions but I'm having difficulty with the need for each view to require a separate sheet. This has a high development and maintenance cost because my customer has many products and for each view they want to develop a different custom data connection. As you can image this means even though they want the same reporting across multiple products that have to duplicate their entire report stack every time they add a product. What's more, any changes to the existing views also have to be replicated across all products.
I can imagine many occasions where a business would segregate their data sources but deliver the same reports/dashboards to multiple customers/products/etc. Given this, they would want to develop one view/dashboard and swap out the reporting with a click.
I've created a sample workbook to illustrate the complexity and the issues with parameters and actions. And I opened support case but Tableau doesn't current support swapping out entire data sources. Creating a single template workbook and then Replacing Data Sources may be an option but in the real-world workbooks there are already ~10 data sources supporting 10 views and more sodas than coke and pepsi. This means there's extra complexity and maintenance with this approach.
I hope someone has encountered this before and has some suggestions. Thanks in advance for your help.