This time we’re going bigger than ever. Fabric, Power BI, SQL, AI and more. We're covering it all. You won't want to miss it.
Learn moreDid you hear? There's a new SQL AI Developer certification (DP-800). Start preparing now and be one of the first to get certified. Register now
We have rolled out a new high-level summary metrics experience in the Power BI Admin Portal . This new experience replaces the four summary tiles you previously saw : CPU, Memory thrashing, Memory usage and DirectQuery with wide range of metrics measuring the summary of usage in the last 7 days to better portray the health of your capacity. This health experience supports capacity admins to make data driven decision without having to leave the portal.
New_Power_BI_Premium_summary_and_workload_metrics_available_in_the_admin_portal
Below the overall summary metrics, usage is split between different workloads enabled on the capacity which can be expanded upon by clicking on the downward chevron on the right. The colored bar represents the amount used out of the total metrics available for the given workload.
New_Power_BI_Premium_summary_and_workload_metrics_available_in_the_admin_portal
As per the screenshot above in the past 7 days : Dataset workloads used 4.42% of CPU and 8.72 GB of memory on an average. A total of 180 datasets were refreshed of which 1 had failed. Average duration of refreshes was 6 minutes and 38 seconds with an average wait time of 4 seconds. A refresh will wait when there are not enough resources (memory or CPU) available for it to start. These resources may be consumed by other refreshes or queries being run on the capacity. In the past 7 days , there were dataset(s) that took over 44 mins to refresh and waited for 67 minutes to start refresh. In this capacity, a total of 7.1 interactions were made on the report with an average response time of 4 seconds and 2 ms wait time. A total of 51 datasets were evicted. When a capacity faces memory pressure, the capacity node evicts datasets that are not in use to accommodate new requests.
New_Power_BI_Premium_summary_and_workload_metrics_available_in_the_admin_portal
As shown in the summary above, for the past 7 days a total of 14 dataflows refreshed were requested, of which 13 succeeded and on an average run for 1 minute with zero wait time. There was no lag between the scheduled time and the start of the dataflow refresh.
New_Power_BI_Premium_summary_and_workload_metrics_available_in_the_admin_portal
These metrics (above), along with the system metrics, give a summarized view into the health of a capacity. Long refresh wait times, can be a sign that a capacity is becoming busy. A refresh will wait when there are not enough resources (memory or CPU) available for it to start. These resources may be consumed by other refreshes or queries being run on the capacity. Long query durations could be a sign that the capacity is busy. It may also mean that a single dataset is causing problems and further investigation is needed.
There are times when you want to access the raw data and do your own analysis. Instead of exporting the underlying data separately by each metric, you can now easily export all the underlying aggregated metrics in one go from the Power BI Premium Capacity Metrics App. You can also export from the visuals themselves, or use the Analyze in Excel by clicking on the ellipses menu on top right of the report, and access PivotTables, chart and slicer features in Excel.
To install the Power BI Premium Capacity Metrics app, click here
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.