Prevent refresh of a table in Table.combine

[current_thing]

Hello,

I have created a PBI file that consists of:

Book1 will consist of data that will not change. Book2 will consist of data that will change. Query2 is a union of Book1 and Book2.

Goals:

1) In Desktop view, the user will see only Query2 which I've accomplished

2) When I refresh Query2, I only want the refresh of Book2 since this will be dynamic data. How can I accomplish this? Even though I specifically set option to not refresh Book1, it still does when I refresh Query2

Hope my question is easy to understand with screenshots. Thanks!

[lbendlin]

Even though I specifically set option to not refresh Book1, it still does when I refresh Query2

That is by design.  The "do not refresh"  request will be ignored for queries that other queries depend upon.

[current_thing]

Thanks for replying. Can you think of a setup that'll accomplish what I'm trying? The static dataset is HUGE, so it is cumbersome and waste of resource to refresh it.

[lbendlin]

Consider using incremental refresh or other manual partition management processes.

[current_thing]

I believe incremental refresh requires premium subscription, which my company won't pay for. Can you tell me more about manual partition management? Does this require premium subscription? Thanks.

[lbendlin]

yes, it requires access through the XMLA endpoint which is a premium feature too.

That means your only choice is to do your own pseudo partition management (ie CSV files on a OneDrive) and accept the cost of loading all these files all the time.  

[current_thing]

My other idea was to convert the static csv import into permanent table. Meaning that it will no longer be connected to CSV file. Then using DAX expression of Union, I can combine the pemanent table and static CSV file. Do you have suggestions how to go about converting CSV table into permanent table? Thanks!

[lbendlin]

that's a bad idea.  The DAX UNION will double the amount of memory your semantic model needs, for no good reason.  Don't do that.

Static CSVs are the least bad option.