Do you mean the difference between Data blending and joining?
Many forum threads out there as well: http://community.tableau.com/thread/117556?start=0&tstart=0
As a general guideline, if all your tables are from the same source, joining is ideal in this scenario. Data blending is often used to join data from different data sources. This is not set in stone, though, and once you understand data blending more, you can decide which will work best. Another often cited example for data blending is when you have summary information and want to link to detail information.
Per "Deciding Between Joining..." article above:
If a workbook uses two data connections from the same data source, joining the data tables can improve performance and filtering control. If you are adding data to an existing data source or a table in the same data connection, using a join is better.
However, within these guidelines, sometimes joining data tables may not work as well as blending data. Common situations that may perform better with data blending include the following:
- The data source contains too many records for a join to be practical.
- You want to display a summary and details at the same time.
You said "Is there a way to improve performance when doing calculation on the tables"; what types of calculations? Do you have a packaged workbook example you can post?
To me, it sounds like joining your tables together would be ideal -- but I don't know anything about the data, either, so I could be wrong. We'd need more information about the data in the tables to help you make a decision.
I am having problems with just adding a dimension. I currently have filters to hopefully speed up but it is currenly taking over 15 min just to add a year dimension. Is there a way to change from linking a table to joining for the existing table, or do I have to delete the existing table and add a new data connection? The data tables have millions of rows but i only need to analyse about 5,000.
The best alternative, considering you have access to manage the database, is to create a view or a physical table in the database, based on the relationship among the tables that you need. Then using Tableau connect to that new table.
There are many, many potential reasons why your workbook could be operating slowly--are you using an extract? You can set data source filters to only return X amount of years (sounds like you have several years worth of data if it takes 15 mins to add a Year dimension). There are many other performance tips out there as well--but extracting the data and limiting what is brought into Tableau may be most helpful in this instance.
In order to change to joining the tables, you'll have to create a new data source. However, if you've built out some stuff already, you can try creating the new source, and then replacing the old data source with the new.
PS--Ramon is correct, if you can create a view with all your joins defined in it, it will be the most optimal scenario.