top of page
Jihwan Kim

Contrasting Semantic Model Direct Downloads and Azure DevOps Integration in Power BI

In this writing, I'd like to share my insights into the differences observed when configuring deployment rules in test and production pipelines. Specifically, I'll compare the experience of downloading directly from each workspace to a local computer versus synchronizing with Azure DevOps and then cloning to the local machine. While this scenario may not commonly occur in real-world situations, I aim to highlight the variances I have discovered.


Let's start with an assumption:

1. Download directly from Dev workspace:

- Outcome: Successfully download the pbix file, able to open and view the report and semantic model.

- Next steps: Deploy to test and prod, then configure deployment rules in Test and Prod workspaces.

Now, when attempting to download directly from Test or Prod workspace:

- Outcome: Successfully download the pbix file, but unable to view the report and semantic model.


This indicates a difference – the semantic model in the Test or Prod workspace doesn't display anything when downloaded directly. I'll verify whether this holds true.


Now, let's explore another assumption:

2. Sync Dev workspace with Azure DevOps and then clone to the local computer:

- Outcome: A common and practical process for many developers, resulting in the ability to see both the report and semantic model on the local machine.

- Next steps: Deploy to test and prod, then configure deployment rules in Test and Prod workspaces.

Subsequently, when attempting to sync Test workspace with Azure DevOps and clone it to the local computer:

- I'll investigate and share my findings.



Let's start.


I generate a report with two parameters as follows.


In the Power Query Editor, two parameters are established to enable developers to focus on a specific subset of countries and products, allowing them to work with a manageable amount of data on their local machines.


I integrated these two parameters into the country dimension table, product dimension table, and the sales fact table as illustrated below.


I published it to the Development workspace, and as anticipated, it displays two countries and two products.


From the Development workspace, I downloaded the semantic model and it was a form of a PBIX file. Upon opening the file, it correctly displays all the information as shown below.



I transferred the report and semantic model from Dev to Test, then set up deployment rules in the Test pipeline to include all countries. The report in the Test workspace displays as anticipated.




Upon directly downloading the semantic model from the test workspace and opening the PBIX file, the display is as follows.





Deploying it to the production workspace and configuring deployment rules to display all products, the report in the prod workspace showcases all countries and products.





As anticipated, downloading directly from the production workspace yields the same results as when downloaded directly from the test workspace.





As illustrated earlier, downloading directly from the Test or Production workspace does not display any visualizations or semantic models.



Now, let's explore the results of synchronizing the Test workspace with Azure DevOps and cloning it to the local computer. I omit showcasing the outcome of synchronizing the Dev workspace with Azure DevOps and cloning it to the local computer since understanding the results from the Test and Production workspace experiments allows us to easily infer the Dev workspace's outcome.








As a result, all visualizations and the semantic model are correctly displayed. Surprisingly, the parameter setup is also identical to that of the Test workspace.




Executing the identical procedure in the Production workspace produces the anticipated results, highlighting a consistent parameter setup. The process, involving synchronization of the Production workspace with Azure DevOps and subsequent cloning from Azure DevOps to the local computer, reveals all countries and all products.





In this exploration of deployment procedures in Power BI Service, I delved into the nuances of configuring deployment rules in the context of different workspaces—Dev, Test, and Production. The primary focus was on comparing two approaches: direct semantic model downloads and integration with Azure DevOps.


Crucial insights gleaned from the examination reveal that direct downloads from Test or Production workspaces do not exhibit visualizations or semantic models. Conversely, synchronizing with Azure DevOps and cloning to the local machine not only reveals the visualizations and the semantic model but also mirrors the exact setup of parameters observed in the Test or Production workspace.


Though the likelihood of downloading semantic models directly from test or production workspaces and encountering empty PBIX files is rare, I aspire for this information to enhance your positive learning experience.










183 views0 comments

Comments


bottom of page