Horizontal scaling of the semantic model

[[email protected]]

With SCC, you can save time and effort by maintaining data connections and credentials. You can create an SCC for a cloud data source and share it with other Power BI users who have Use permission. These users can use the SCC to connect your semantic models and paginated reports to the data source without needing to know the credentials, which are protected by the SCC. You can also create multiple SCCs for the same data source if you want to use different connection options—such as different credentials, privacy levels, or single sign-on modes—for different Power BI artifacts.

You can now enjoy the new semantic model scale-out feature, which enables you to have more efficient and scalable production solutions. With this feature, Power BI automatically creates read-only saudi arabia telegram mobile phone number list replicas of your semantic models, both in Import mode and Direct Lake, so you can serve more users simultaneously without losing performance. Additionally, read-only replicas are not affected by refresh operations you perform on the read/write replica, giving you more flexibility and isolation for your data refresh scenarios.

Here are some of the advantages that horizontal scaling of the semantic model offers you:

Better query performance: Power BI automatically adjusts the number of read-only replicas based on query volume and variability.
Update isolation: Update and write operations that you perform on the read-write replica do not affect query performance on the read-only replicas.
More flexibility for advanced data refresh scenarios: You can disable automatic replica synchronization, perform all the refresh operations you need on the read/write replica, and then manually synchronize the read-only replicas.
Semantic model horizontal scaling is one of the key features that make Microsoft Fabric and Power BI superior to Azure Analysis Services (AAS) as they automatically scale to live user demand, without the need for prior analysis, scripting, or monitoring. Additionally, the cost in Fabric depends on usage, while in AAS it depends on the number of replicas.