There are advantages
to using a published data source, but there is one downside to using them. As mentioned in this link,
it is not possible to edit a published data source directly on Tableau Server
and requires that the developer make a local copy of the data source. The
method described in the link is not ideal for the following reasons:
- For large data sources, it can take a lot of time to download the data sources to your local machine and republish them to the server
- Replacing data sources can reset quick filter formats on dashboards (e.g. converting single value dropdowns into multi value lists)
- Since this method is time consuming, it makes it difficult to quickly make changes based on feedback and adapt to changing business requirements
The method that I’ve been using to edit published data
sources is able to prevent all of these issues. The basic idea is that you
don’t need all of the data to edit published data sources. You just need the
metadata. After you have created an embedded extract using a limited amount of
data in Tableau Desktop and built your dashboards/visualizations, follow these
steps:
1. If you have a large data source, create an empty extract in Tableau Desktop and publish the empty extract to Tableau Server.
After it is published, you can refresh it manually on the server so that the
full data source will be populated on Tableau Server instead of your local machine.
3. Replace the data source in the workbook from the embedded extract in the workbook to the published data source on Tableau Server. Verify that nothing breaks or reformats when you make this replacement.
4. Do not delete the embedded extract. Instead, re-extract the embedded extract with only a subset of the full dataset (e.g. 5 sample records or 1 day of data). (This is explained later).
5. Publish the workbook to the server with the
dashboards/visualizations connected to the published data source and the
embedded extract disconnected from any dashboards/visualizations.
6. Whenever you need to make changes to the published data
source on Tableau Server, make the changes in the embedded extract in the
workbook, create an empty extract of the embedded extract if necessary, and
then publish the embedded extract with the changes to the server overwriting the existing
published data source.
The reason why you are maintaining the embedded extract is
so that you will still have access to the metadata of the data source without
having to download the data source from the server. Having a few rows in the
embedded extract allows you to test the new calculations or changes on a small
amount of data and reduces the size of the workbook so that it can be
downloaded or published to the server quickly. When the published data source
is overwritten, the changes will be reflected in all of the
dashboards/visualizations connected to the published data source without having
to use “Replace data source.” If you need to, you can right click on the
published data source in Tableau Desktop to refresh the connection.
Now that you know how to work with data sources, check out my post of How to Default your Tableau Dashboard to the latest date of a data source.