Organizations spend lots of money ensuring they have the right resources to support their Power BI platform to ensure a good developer and end-user experience. This is especially true if your organization has purchased its own Premium Capacity. Because of this large investment, you’ll want to make sure that you have what you need to appropriately monitor your organization’s premium capacities.
Enter the Power BI Premium Capacity Metrics App, a canned app designed to give admins the necessary monitoring capabilities to make informed decisions about their Premium Capacity.
In this blog, we’ll explain the PBI Premium Capacity Metrics App, describe how to install it, and provide some tips on how to use it.
What Is The PBI Premium Capacity Metrics App?
The PBI Premium Capacity Metrics App is a template app, meaning it’s publicly available to use within your organization’s tenant. This app equips Power BI admins to be able to monitor their capacity’s resources and identify issues and areas for improvement.
Let’s start with an example of how you, as a Power BI admin, might use this app.
When a dedicated capacity in Power BI reaches its maximum capacity, meaning its compute resources are fully utilized, the admin team will receive an email notifying them and explaining that their capacity will be throttled. This will impact the dataset and report performance, thus negatively impacting user experience.
As an admin, you can use the app to identify where the issue came from, the exact point in time that it occurred, and even the user that caused it so that you can make them aware and provide mitigation tips.
As previously stated, the PBI Premium Capacity Metrics App is a template app and is available in the AppSource to be installed (instructions for installation are in the next section).
How To Install And Set Up The PBI Premium Capacity Metrics App
Installing the PBI Premium Capacity Metrics App is easy and can be done directly from Power BI Service.
Step 1: Open the “Apps” page using the left panel and click the green “Get apps” button on the top right.
Step 2: Navigate to the “Template apps” section and search Premium Capacity Utilization and click on the app icon (note there is still a previous version available, make sure you have the one in the screenshot below).
Step 3: In the next two pop-up windows, click the blue “Get It Now” button.
Step 4: Click the “Install” button in the next window (this step may take a few minutes).
Installing this app will automatically create a new workspace with the app already installed in it; there is no alternative to this and you will not be able to install it into an existing workspace.
Step 5: Open the newly created workspace which will follow this naming convention in the screenshot below. Click the “Connect your data” button on the yellow banner towards the top.
Step 6: Fill out the information in the pop-up window accordingly. If you have multiple capacities, you will NOT need multiple apps, just use one of the ID’s, and it will automatically connect to the others in the same tenant.
Step 7: Follow the prompts on the next screen and click the “Sign in and connect” button.
Step 8: Add the app to your list of favorites by going back to the “Apps” page using the left panel, then clicking on the “Get apps” button and searching for the Premium Capacity Utilization and Metrics App.
How To Use The PBI Premium Capacity Metrics App For Performance Monitoring
Using the PBI Premium Capacity Metrics App can be daunting because of the amount of information it can provide. For that reason, we will use this section to introduce the different components of the app and provide some advice on where a user’s time is best spent.
The app has five different pages that admins can use to help monitor their capacity and triage any issues that may come up, which are described in more detail below.
At the top of each page is a slicer for the user to select which capacity they would like to view.
Overview Page
The Overview page provides high-level details of the overall performance of the capacity and has three sections.
This section is the left portion of the page and consists of two separate visuals stacked on top of each other.
Drive your focus to the matrix on the bottom left, which displays metrics (described in the table below) for each Power BI item in the selected capacity. Use this matrix to highlight problem artifacts that could be draining your capacity’s resources. And expand the individual artifact to view what kind of operation could be causing the problem.
Performance
The performance section is to the right of the Artifacts section and has four visuals. This section is perhaps the most important and will help the user pinpoint exactly when, where, and who caused a capacity to reach its max threshold.
CPU over time: visual on the top displays CPU usage for the capacity broken out by background and interactive activities. Background operations are activities that are not directly triggered by users, such as data refreshes. Whereas, interactive operations are caused by users, such as loading pages or filtering.
Right-click to drill through any part of the visual to open the Timepoint Details page to view exactly what background and interactive operations caused a spike at that given moment.
Overloaded minutes per hour: displays a value showing how severe an overload was on the performance of an artifact.
Item size: shows the memory recorded for the Power BI items in the capacity over time.
Performance profile: displays report performance broken out into three different categories (Fast/Moderate/Slow). The performance aggregation is taken from the number of operations conducted on an artifact.
Weekly Trendlines
This section has four sparklines that provide quick, high-level insights on the trends for each of the following data points included.
Evidence Page
The Evidence page of the app helps to identify which Power BI artifacts are leading to overload and which items are impacted by that overload.
Overloaded Item: the most important visual here is the table on the bottom left, which displays Overloaders and Overloaded Items. The overloaders highlight items that cause highly impactful overloaded events. Toggle this visual to Overloaded Artifacts, which displays items most impacted by overloading in the past two weeks. Users can drill through the individual artifacts to get more information specific to that artifact.
Overloading windows: helps users understand if an overload or autoscale events occurred because of a single artifact or several.
Items overloaded (seconds)use this visual to understand if items that are overloading impact their own performance, or if they impact the performance of other items
Number of users overloaded: use this visual to understand if a single user was impacted by an overload event, or several users. This helps the user understand how widespread of an issue this is.
Refresh Page
This entire page is dedicated to helping users identify refresh performance issues and can be accessed for an entire capacity or as a drill-through for specific items or. The top of the page allows users to make selections that impact all of the visuals on the page.
Refresh by item: shows the breakdown of items by the metric filtered to (using the slicer described above). This visual is helpful when trying to understand which optimizations for an item can help reduce the capacity footprint.
CPU: the columns in this visual represent the number of CPU seconds used to finish a single operation per hour across a fourteen-day period.
Duration and Operations visuals: the columns in these visuals show the number of seconds a single operation took to finish a single operation per hour across a fourteen-day period.
Refresh Detail: his matrix shows the metadata for all of the individual refresh operations that occurred. The ratio column should be used to help understand the ratio between CPU time and processing time. A low ratio means there are inefficiencies that cause Power BI to spend more time waiting for the data source and less time executing the refresh.
Timepoint Details Page
This page is perhaps the most important page when determining the causal factors that lead to overload or your dedicated capacity reaching its maximum threshold. Access it by drilling through various points in other pages to view the operations that occurred at that time.
The page is broken into two primary sections. There is a table for interactive operations and a table for background operations. As described before, interactive operations are activities performed by users at the front end, such as loading pages, making visualization selections, and filtering.
Background operations are not caused directly but users and are operations such as dataset refreshes. Each of the tables displays the exact artifact, the operation, the start and end time, whether or not the operation was a success as well as helpful metrics that describe it’s impact to the capacity at that point in time.
I recommend using this view when you want to know the specific operation that caused an overload and would like to contact the user.
Item Details Page
The Item Details page is another drill-through page that offers very low-level information specific to a single item that the user drills through. This page displays the information provided for the past two weeks regarding CPU usage, Users, Duration, Overloading, and more.
Closing
Organizations spend lots of time, money, and resources ensuring their teams have the necessary items to report on their business metrics efficiently. If your organization has licensed Power BI for a dedicated Premium Capacity, you should take the necessary steps to monitor and act proactively to find efficiencies that will lead to better and faster insights, and thus a better user experience leading to greater adoption.
The PBI Premium Capacity Metrics App is an excellent tool that helps organizations do that, and since it’s a template app, it requires no effort to develop and minimal effort to install.
If you have more questions about the PBI Premium Capacity Metrics App or Power BI in general, our experienced Power Platform experts will gladly guide you.