15. November 2023 By Mark Hartmann
Try MS Fabric as a Pilot Project - No Reservations, no Risk
Microsoft Fabric: Potentially revolutionary and still a work in progress – check it out now?
Microsoft Fabric has the potential to revolutionize the way you build, extend, and maintain enterprise data architectures. Years ago, the concept of "Platform as a Service” (PaaS) rather than "Infrastructure as a Service” (IaaS), brought a simplification to building architectures. With Fabric, Microsoft now offers the next, logical step: Turnkey "Software as a Service” (SaaS) modules can be assembled with a few clicks into a powerful, easily extensible, and scalable enterprise data architecture. There is no longer any need to deal with individual platform components.
This sounds like the future, and it is. However, this future is still in the making: Fabric has been in "Public Preview" since May 2023 and in "General Availability" since 15th November 23. With the help of selected partners like adesso, Microsoft has been able to implement an impressive range of Fabric features since the end of 2022. A look at the roadmap (Microsoft Fabric release plan documentation - Microsoft Fabric | Microsoft Learn) shows what will be offered in terms of further development in the coming months.
Business Stakeholders: Different perspectives and individual desires - but the same reservations
You are a stakeholder of a company that relies on cloud components and services. What are you going to do? Wait a few years to see how Fabric evolves? If you do, you may miss a golden opportunity to get in on the ground floor of the future. Are you going all in on Fabric now? Then you risk getting stuck with the limitations of current development.
Maybe you're a CTO and want to make a fundamental decision about new technology. Maybe you oversee the IT department and want to evaluate new approaches. Maybe you are focused on content-driven, self-service business intelligence and are interested in new possibilities.
All of these different perspectives are united by the same idea: a pilot evaluation of Microsoft Fabric must be done without risk and at little or no cost. A situation must be created in which all reservations about early engagement with Fabric are removed early on. This can be achieved through an A/B comparison of implementations.
A/B comparison of implementations: No risk, low (or no) cost
Setting up a pilot using Fabric as an A/B comparison of two alternative implementations can be a useful way to conduct a detailed and critical evaluation of Fabric without risk and at little or no expense:
Two implementations of the same project are created: a classic variant "A" (without Fabric) and a Fabric variant "B". Both implementations are based on the same business and professional requirements. However, the technical implementations differ significantly.
Business Intelligence Reporting as an example. The basic requirements
To make an A/B comparison of two implementations, it is first necessary to clarify the requirements from different perspectives. For instance:
- Initial situation: Data for a specific business scenario is stored in an Azure SQL database and should be prepared using an enterprise data architecture and made available for self-service business intelligence purposes such as reporting.
- Business requirements: Power BI reporting for the business scenario is to be realized based on this data.
- Business requirements: A common data architecture should be used so that the data for this scenario can be used for other use cases in the future.
- Technical / IT requirements: Implement this data architecture for the scenario should be carried out with Microsoft Cloud components.
Option A: Classic implementation with Power BI Premium Cloud Service & Power BI Reports
- An existing Azure SQL Database is connected to the Power BI Cloud Service via a connector.
- Data from the existing Azure SQL is loaded into the Power BI Cloud Service via Power BI Dataflows.
- Power BI Dataflows transforms the structure of the data or tables to create dimension and fact tables for a star schema data model.
- Dimension and fact tables are assembled into a star schema data model via XMLA End Point in the Power BI Cloud Service.
- Processed data is regularly updated to the Power BI Cloud Service using import mode, where it is made available in a dataset for Power BI Reports.
- The Power BI Reports access the data from the dataset and visualizes it according to business specifications.
Option B: New implementation with Fabric Cloud Service & Power BI Reports
- An existing Azure SQL Database is connected to the Fabric Cloud Service via a connector.
- Data from the existing Azure SQL is made available in the cloud service via Fabric Shortcut or Fabric Data Factory.
- The structure of the data or tables is transformed using Fabric Data Warehouse, to create dimension and fact tables for a star schema data model.
- Dimension and fact tables are assembled into a star schema data model via XMLA End Point in the Fabric Cloud Service.
- The prepared data is available through Direct Lake in the Fabric service, where it is provided as a dataset for Power BI Reports.
- The Power BI Reports access the data from the dataset and visualize it according to business specifications.
Comparing the two variants: Advantages for Fabric
|Implementation Topic||Classic Variant "A"||Fabric Variant "B"||Comment|
|Connecting the Data Source||Connector for Cloud Service||Connector for Cloud Service||Identical in both variants|
|Load or provide data from source||Load or provide data from source Power BI Dataflows in the Cloud Service||Fabric Shortcut or Data Factory in the Cloud Service||Fabric One Lake as the universal basis of data|
|Load or provide data from source||Power BI Dataflows in the Cloud Service||Fabric Data Warehouse in the Cloud Service||Fabric One Lake as the universal basis of data|
|Structure of the data model from tables||Via XMLA End Point in the Cloud Service||Via XMLA End Point in the Cloud Service||Identical in both variants|
|Providing Data for Reports||Via Power BI Import as Dataset||Per Fabric Direct Lake as a dataset||Fabric Direct Lake eliminates the need for another copy of the data|
|Data in Reports Visualizers||Per Power BI Reports||Per Power BI Reports||Identical in both variants|
|Professional Requirements||Fully Fulfilled||Fully Fulfilled||Fully fulfilled in both variants|
|Business Requirements||Partially fulfilled||Fully Fulfilled||Fully fulfilled only in Fabric variant|
Evaluation of both variants: Advantages for Fabric
The example shows that both the classic variant "A" and the Fabric variant "B" meet the professional requirements equally well. However, the technical implementation differs in key points:
The Fabric variant "B" provides the data via direct lake, while the classic variant "A" has to reload it via import. This means that variant "A" creates an additional copy of the data, which Fabric avoids in variant "B" – a potentially decisive advantage.
The loading/provisioning/transforming of the data or tables is done in the classic variant "A" with Power BI Dataflows. The Fabric variant “B” uses Data Factory and Data Warehouse. This means that the data in variant "A" is "restricted" to the Power BI cosmos and can only be used to a limited extent for other use cases. The Fabric variant "B", on the other hand, is open to other use cases.
In this example, the result of the comparison would show that only the Fabric implementation meets the business requirements, while the classic variant does not. By comparing the two implementations, a targeted and critical evaluation of Fabric is possible that can be applied to all relevant individual goals. Fundamental aspects such as OneLake as a universal database or, in the future, Purview for overarching governance complete the picture of Fabric.
Implementation and financing: Commission one variant, get both
The previously described variant "A" (classic implementation, without Fabric) offers security in terms of implementation. Even if Fabric is not yet ready for productive use in a specific case, the technical requirements are met by the classic implementation. Therefore, there is no risk for the implementation and it can be requested as part of a normal order with partners such as adesso.
The additional variant "B" (Fabric Implementation) offers a dedicated and critical comparison with Fabric. In some cases it offers significant technical differences while meeting all other requirements. Through close cooperation with Microsoft, partners like adesso can offer the Fabric variant for free or at low cost.
Provided that the classic implementation is also commissioned, there are no or only low costs for an additional Fabric variant. The reservations against a pilot project can thus be reduced or even eliminated.
Want to know more?
Would you like to learn more about exciting topics from the adesso world? Then take a look at our latest blog posts.