For a long time, we have been using a “Golden Semantic Model” (Golden Dataset) approach for Microsoft Power BI reports when we need to access similar data for multiple reports. Microsoft Fabric justifies this approach and turbo charges it.
What is the Golden Semantic Model?
The Golden Semantic Model is a unified semantic model that integrates various data sources; cleans, transforms, and shapes the data; and extends it with DAX measures. Once published to the Power BI Service, it allows for the creation of lightweight reports with live connections. This method offers several key advantages.
Advantages of the Golden Semantic Model:
- Single Data Connection: By leveraging a single data connection for production systems, you only need to update credentials in one place when they expire, simplifying management.
- Scalability: Adding new reports doesn’t impact production systems as there is no need to repeatedly pull the same data into different models. This scalable approach ensures efficient resource utilisation.
- Improved Data Quality: The model enhances data quality by ensuring consistency and reliability.
- Maintenance: Easier maintenance, as changes are automatically reflected in the connected reports. There is no need to manually update multiple reports.
- Single Refresh: All reports connected to the Golden Semantic Model are refreshed simultaneously, ensuring that everyone has access to the most up-to-date information.
- Consistent DAX Measures: While the core DAX measures remain consistent across all reports, Power BI developers can still add their own custom measures without affecting the central model. This flexibility promotes innovation and customisation.
However, there are some disadvantages to this approach, such as the added complexity of managing reports with varying requirements while ensuring model changes don’t break connected reports. Increased data security and compliance needs must also be taken into consideration. Finally, if the model is large and the reports have a large concurrent audience, performance can be affected due to the shared capacity nature of the Power BI Service.
Why Microsoft Fabric?
Fabric is the unified data platform for end-to-end analytics. It endorses the Golden Semantic Model approach and extends it to address some of its disadvantages.
The Lakehouse plays the same role as our Golden Semantic Model; in fact, the Lakehouse comes with a default semantic model. It also has additional benefits when compared to Golden Semantic Models.
Data can be loaded and transformed using several different tools; we are not limited to using Power Query data flows like in semantic models. We also have data pipelines (data factory) and notebooks (Python and Spark), not to mention Gen 2 Data flows.
Individual tables can be refreshed on their own timetable. This reduces the resource waste that would otherwise be spent refreshing unchanged data as part of a full refresh.
Time travel is possible. We can historically query what the data looked like on a certain day. This provides an extra level of auditing and compliance.
Other workloads like Data Science can easily access and use the data.
OneLake Shortcuts allow data reuse at a whole new table level. We can share an individual table rather than the whole model to be used in different Lake houses. The data is always kept in sync as it uses the One Copy functionality.
Dedicated Capacity Units in Fabric, compared to the Shared Capacity we have used in the past, can resolve performance issues for large audiences by allowing organisations to choose resource levels. Compute capacity can be scaled up and down as needed, with smoothing and burst capacity available to provide optimised usage and consistent performance.
OneSecurity provides hierarchical security controls enabling easier ways to govern and manage. We use partitioned data tables in the Lakehouse in conjunction with row level security to optimise performance and ensure users can only access the data segments they are permitted to, enhancing both efficiency and security.
AI ready with Copilot provides an accelerated time to insights. We can leverage conversational language to preform many of the tasks like creating dataflows and pipelines, generating code, building machine learning models, and visualising results with report pages.
The good news is that we don’t need to abandon our Golden Semantic Models. We have the choice to continue to use them without change, or we can convert them to a Lakehouse and take advantage of the new wider data analytics platform tools we have with Fabric.
Let us know if you want to explore how Professional Advantage and Microsoft Fabric can modernise your data and analytics needs.
Are you looking to evolve your data strategy?
Schedule a free 90-minute consultation with us today.
Address your current data challenges and lay the foundation for leveraging data as a strategic asset for the long term.
