Publishing data sources to Tableau Server & Tableau Cloud: Tableau Tutorial for Beginners: 

Love that you found the newer video! To be honest published data sources are still a bit frustrating but Tableau have this guide here thats i think summarises the best ways to use it. help.tableau.com/current/pro/desktop/en-us/publish_datasources_about.htm

That not really a server topic. More of a dashboard design topic but I'll add that to the list of content. The simple answer is used a fixed dashboard size and make sure that when people view your dashboard they have to scroll as little as possible which usually means 1200 by 800 for width and height.

Thank you Tim ..Sure.I will follow your guidance on this.

Workbooks will send their query through the server data engine which then looks to see if the query can be answered by cache and also checks to see if the data freshness policy for the data source requires it to be update the cahce. If the cache needs to be cleared it then send the request to the database and if not it sends the query results back to the server VIZQL component which then renders the chart ( if that has no cache ) and you see the chart. I'll maybe do a video on this process and caching in general.

You need the data management add on for the lineage to work fully. See the second paragraph on this help article. help.tableau.com/current/server/en-us/dm_lineage.htm

I’m a more recent update people with access can edit calculations in published data sources. I’ll find the link and add it here >> Edit published data sources: New in Tableau 2021.4 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-tnIc0eIoIvI.html

not with a csv attachment but a pdf or image attachment yes. Its possible to go to the dashboard and download the csv but emailing out csvs causes most admins security problems. There are other tools that do this.

You can't loose data. A published data source always has a reference to the origin so really that can only happen if you delete the original flat-file that created the published data source. even then you can export the data from the published data source as an admin and recreate the original data that was in the flat file. All other data sources are databases and the published data source members this in its metadata which I hope are harder to delete. At the end o the day you cant protect data from irresponsible users.

It's not dumb at all. really good question. I'll keep it simple for the comments but the first thing to do is find any downstream assets. You can use the metadata API to do this, this will show you workbooks or views using the connection. Switch out those connections for your new data source and then once you've cleared it, I would actually archive it next into a folder for a month or two just in case it's linked to something else but you're not aware. A month later once you're happy nobody is using it delete it. BUt it's not a simple as that as what we're really talking about is metadata rather than a physical object so it starts to become more about governance approach. Great video suggestion though.

@@TableauTim Thank you for your reply, way above my current skill level but very helpful.

No limit as many as you can build and maintain.

What version of Tableau are you on at present? You should be able to publish up a data model as a published connection. If you have multiple sepearte connections just repeat the process for each one or am i misunderstanding your question?

@@TableauTim I am on the most up to date. I believe that did work, I had it published as a extract + multiple connections. The only issue is I had it deleted off tableau online. I do have the local copy but both are saying to edit the connection but not allowing me to connect to anything. Is there a way to remedy this?? Thank you so much!!

Analysts. Admins in my experience become the bottle neck and also don’t know which data sources have value. Only exception are really small business where the admin is also the author and also the analyst. The other way to think of it is that analysts who are publishing data sources are enabling other analysts to go off an build more content. What this really means is you have a hierarchy of analysts some who just consume who just use reports, analysts who consume that can also use published data sources to self serve with their explorer license building charts from existing data sources, analysts who consume and build who have creator licenses and data pipeline engineers who don’t consume but also publish data sources that all the previous user use. The mix of what you have depends but none of those people are server administrators. Their job is to keep the server running, manage governance and security and make sure no monsters are growing legs on the systems.