Before we jump into creating and using a data flow, I briefly want to discuss why you would use one. The reasons are twofold. A data flow is a series of transformations, as in ETL processes, that result in a dataset designed for reporting. This results in organizational-wide timesaving as report authors don't have to reinvent the wheel by performing the same transformations on the original data. It also means that the transformations are consistent as the data flow ensures one version of the truth. A data flow pulls together information from various sources, puts it through the necessary transformations, and makes the end product of the flow available through Azure data Lake gen 2 storage.

Let's go through a simple example of creating a data flow that ingests an Excel workbook of sales targets from Azure blob storage and splits it into two gross sales and margin targets tables. We set up a data flow through the Power BI.com portal. I'm going to create the data flow in my GettingData workspace. Go to your workspace and create a new data flow. I'm going to start by adding new tables. As with Power BI desktop, you have a vast array of data source options because we are using online Power Query editor. 

My target data source, excuse the pun, is stored in an Azure blob container. After entering the storage account name, I'll choose account key as my authentication method and paste in the key from the Azure portal. After clicking next, I'm presented with a typical data import navigation window. Having selected the Branchtargets container, all the files are stored within are displayed with the metadata. This window is for selecting a container within your storage account, and as I have only one container, there is nothing to do here except click transform data. The next window shows just the Branchtargets container in the Power Query editor that we are familiar with. If you have lots of files in your container, you can use column filtering, like the file extension, to select the files you're interested in. 

I'm only interested in the first file, so I'll access its contents by clicking on the binary value within the content column. We are not quite there yet, as now we have to select the sheet within the workbook. There is only one sheet in this case, so that's easy. There is one small gotcha here. The data within the Excel sheet needs to be defined as a table, which involves selecting the data in Excel and using the control + T shortcut. Clicking on table within the data column finally gets us what we are looking for. I'm going to create two tables by duplicating branch targets and renaming it to SalesTargets and renaming the copy to MarginTargets. Next, I'll filter out the rows I don't want for each target type. Margin targets target type is M, and sales targets target type is S. In each case, I'll use the advanced option of the filter row's function. Let's save and close and give the data flow a name.

Now that we've got a data flow, let's use it as a source within Power BI desktop. Under get data, I'll select Power BI data flows. I'm prompted to authenticate before connecting, which I'll do. The data source navigator displays workspaces containing data flows, of which I only have one - GettingData. Within that, there is the Targets data flow, containing margin and sales targets. From here on in, you can treat the table as you would from any other source with the caveat that data retrieval is comparatively sluggish and slow. Well, it was for me.