1 of 1 people found this helpful
Without knowing the size of the data and the volume/type of joins required, it is difficult to suggest what may be best. Generally, with several tables and a significant amount of data, it may be better to de-normalize the data and store it in as few tables as possible; and then use these materialized views, cubes, etc. when using Tableau to create data sources.
If your situation is not too complex, you can create all the joins needed from your existing schema structure and test performance. Do you have a development Tableau Server for testing?
Does your data connection need to be live, or can you publish a .tde, Tableau Data Extract? If the data does not need to be real time, performance with extracts will typically be better and reduce the load on your SQL Server data warehouse. if this is possible, you could create the extract and schedule it to refresh from your warehouse as needed.
Below is a link highlighting best practices when creating extracts such as using an extract filter, aggregating data if possible, hiding unused fields, etc.:
I hope this helps.
THanks a lot for your elaborate response. This is certainly helpful and shows me way forward.