We’re excited to announce a platform update to simplify the way customers manage Power BI Premium capacities. Starting on December 11th 2022, we’ll be rolling out changes to unify the concepts of front-end and back-end virtual cores for capacity management. Unification of these concepts addresses the feedback we’ve heard from customers who sometimes misinterpret the concepts of customer-managed v-cores from Microsoft managed v-cores. This change will simplify the model used by customers to both manage capacities and rationalize usage metrics against capacity sizing decisions.
How do Premium capacity v-cores work today?
Power BI Premium has historically used v-cores as the unit of measure that customers use to reserve compute resources when purchasing a capacity. After a capacity is purchased, its resources can then be used to run a variety of Power BI workloads. Before this change, purchased v-cores in a reserved capacity would be split evenly between front-end cores used by Microsoft for service delivery and back-end cores used by customers to power workload execution. In this model, capacity Administrators only manage back-end v-cores which account for 50% of the purchased v-core capacity.
How are we simplifying our approach to capacities?
The Capacities Team is on a mission to deliver scalable cloud compute with the simplicity of SaaS experiences to our customers. Unification of front-end and backend v-cores for capacity management is another step in the platform evolution to simplify capacity management while offering nearly limitless scale and flexibility. With this platform update, capacity Administrators can purchase and manage resources using a single v-core pool for each capacity.
What does this change mean for my existing capacities?
In order to make this change as seamless as possible for customers, we are enacting changes to both the capacity platform and usage reporting experiences. The amount of effective “throughput” on each capacity SKU will remain the same before and after these changes. Updates will be made to scale usage reporting and adjust for the unification of v-cores.
To better describe the implications of the change we’ll examine a before and after example.
Before unified v-cores:
A customer who purchases a P1 capacity is provided with a total of 8 v-cores evenly split between front-end and back-end services. A P1 capacity would provide 120 seconds of compute over 4 backend v-cores for each 30 second evaluation period. In this example, we will use a hypothetical workload “W1” which consumes 60 seconds of compute. When comparing workload consumption to available compute on the P1 capacity, the platform would measure “W1” as consuming 50% of the P1 capacity throughput during that workload evaluation period.
After unified v-cores:
A P1 capacity provides 8 v-cores of unified compute resources. The 8 v-cores of unified compute will yield 240 seconds of computing resources per 30 second evaluation window. To scale usage proportionally in the new unified v-core model, “W1” will now consume 120 seconds of compute. In the unified v-core model “W1” would still consume 50% of the P1 capacity throughput.
Where will I see these changes?
Customers will see updates throughout the premium product and documentation to reflect the impact of unified v-cores. In the Power BI Premium Capacity SKU table below, you can easily review the before and after impact of these changes.
Capacity Administrators will see these platform changes reflected in the Capacity Metrics Application. Timepoint drill views in the Metrics App will now show:
- Updated values for available compute on each SKU
- Adjusted usage reporting metrics
The percentage of capacity utilization for a given workload will not change because of this update.
How can I learn more about capacity management in Power BI Premium Gen2?
Our documentation is a great first step to learn more about how you can use capacities to power Premium workload engines at scale. For a video walkthrough of Power BI Premium Gen2, you can view our Premium Gen2 Webinar. This includes a detailed walkthrough of:
- Gen2 platform architecture and scalability improvements
- The benefits of our updated smoothing policies for interactive and background processes to simplify capacity management
- A deep dive on how workload evaluation occurs in capacities and its impact to auto-scale and throttling
When will I see changes, what do Capacity Administrators need to do?
These updates will start rolling out globally on 12/11. At that time, capacity Administrators should update to the latest Capacity Metrics App version to guarantee the best experience analyzing capacity usage on the unified v-core Power BI Premium platform.