Beginner questions on how PowerBI "imports/copies" data from source to service

[finarne]

Hi,

Apologies in advance if this information is available elsewhere.

I am starting out with PowerBI and have a few (hopefully simple) questions.

Scenario: simple reporting on a small csv file (<100kb)

If I have a small csv file that I want to use for reporting, when I connect to this csv file does PowerBI "import" the entire file contents to the PowerBI.com service?

My understanding is that the quota I have available with say the PowerBI Pro license is 10GB per workspace, if I have a csv file of 100KB will this 100KB be imported in its entirity uncompressed or does PowerBI do some sort of smart compressing to the workspace?

If I want to connect to a much larger data source (say an Azure Table Storage) that is 5GB in size, will the act of connecting to this data source cause PowerBI to start importing the entire data source from Azure Table Storage to the PowerBI service...or does PowerBI only import a "smart" portion of the Azure Table Storage?

I am trying to understand the end-to-end concepts of how data that I want to report on is imported or copied into the PowerBI service. I am trying to understand if I have a large data source will it take some time for it to be copied/imported into PowerBI or is there another concept to how PowerBI "connects" to the data source for reporting.

I am trying to understand how to calculate the amount of space a data source will consume out of the quota a workspace has available.

In terms of data refreshing, how does this work? Does PowerBI have a incremental refresh of data sources, or does it simpy re-import the data source in its entirity again?

Thanks in advance for any guidance you can provide, I am sure some of these questions will prompt other questions or maybe can be simply answered with a link.

Colin

[Greg_Deckler · Accepted solution]

Yes, Power BI generally imports data unless you specify a "live" (Direct Query) mode of connecting. Yes, Power BI super compresses the data using columnar compression. Difficult to estimate how much compression you will get as that depends on the data.