What is our primary use case?
My usual use cases for Microsoft Fabric involve dashboard development, specifically the retail dashboard that I work on, which has data sources coming from multiple different databases like SQL, API data, and operational data. We need to take data from these sources and create the dashboard in Power BI and publish it on Microsoft Fabric so that external users can access it. This is the overall use case.
What is most valuable?
The features and capabilities of Microsoft Fabric that I have found most useful include it being a uniform platform. Earlier, I have worked on Azure Data Factories and other Azure services, where for each, the data ingestion, data transformation, then data loading, and Medallion Architecture required different capacities and services. But now, with Microsoft Fabric, we get everything packaged. You just connect with your data source, the Data Factory, the pipelining and creation of Medallion Architecture for the Bronze, Silver, and Gold layers, and the Power BI. All these are integrated and using a single compute, you can avail all these services, which was not possible earlier. I see a huge shift and benefit that I am sure all other technical or business users will agree on. This is the shift that I see, the benefit of Microsoft Fabric.
The integration with Power BI has influenced my data visualization through its native availability in Microsoft Fabric. It is automatically available; you just need to add a single item. You select the item you want to work with and type Power BI, and you automatically get options to create the Power BI dashboard. Just as we have Power BI Desktop, Microsoft Fabric also provides Power BI dashboard visualization and real-time dashboards that you can create on the web itself. Additionally, we can create these dashboards on the desktop or a normal laptop with a thick application. When ready for publication, we can simply publish these dashboards to Microsoft Fabric, which automatically follows the same cycle of ingestion, leveraging the unified platform of Microsoft Fabric. It is native, so it feels as though Power BI is automatically part of it while doing development.
What needs improvement?
One thing about Microsoft Fabric that I think could be improved is the potential confusion when working with real-time data versus normal batch loading. If you are dealing with real-time databases or data, you need to work with Event Hub and event streaming, but for normal batch processing, you can usually go ahead with the SQL warehouse or the data lakehouse. I recommend that Microsoft unify these processes, so using one source can meet both requirements. If I am working with real-time data and batched data, I should not have to manage two separate systems; instead of creating an Event Hub and a lakehouse, users could simply manage one house to handle both real-time analysis and batch processing, making it simpler and more user-friendly from both business and development perspectives.
For how long have I used the solution?
I have used the solution for about two to three years.
What do I think about the stability of the solution?
Regarding stability and reliability, I trust Microsoft Fabric because it is a Microsoft product that we have been using for ages. There should not be any questions about the stability of Microsoft Fabric, as it bears the heavy stamp of the Microsoft brand. In terms of analytical capabilities, it provides outstanding features; as a developer, architect, and consumer, I see it from all angles. I can easily create dashboards, ingest data with just a few lines of code, and use the drag-and-drop facility for data transformation, creating my dashboard smoothly. The unique aspect of utilizing a single compute for all these features means cost savings. Given that many organizations are adopting Microsoft Fabric, and with numerous job openings in the market for it, I believe its stability is not in question as it is supported by Microsoft.
What do I think about the scalability of the solution?
From a scalability perspective, I did not see any issues with Microsoft Fabric. We were able to connect to all necessary data sources using multiple connectors. I also utilized Databricks, which provides notebooks for writing transformation code. From a scalability point of view, it is quite good; we can connect with various databases and other tools beyond Power BI. It has comprehensive features, all on a single compute, which makes it manageable to load data into a single lake. The scalability is impressive, and it is easy to categorize data and use SQL features.
How are customer service and support?
I have communicated with the technical support of Microsoft Fabric only once. It was quite easy; I did not have much reason to reach out, as most questions I had were answered in their robust community. I sought technical support relating to billing of a service, and I received a response within twenty-four hours, which met my expectations and was quite good.
Which other solutions did I evaluate?
I have evaluated other options besides Microsoft Fabric, including ThoughtSpot and Sisense, along with a couple of other tools that I cannot recall the names of. We did not consider Tableau and MicroStrategy. In total, we assessed six tools: Microsoft Power BI, Qlik Sense, ThoughtSpot, and Sisense, among others. Ultimately, we decided on Qlik Sense since it provides many workarounds and allows significant customization while writing manual SQLs. My recommendation was to choose Microsoft Fabric, but based on pricing, workarounds, and the essential features we were looking for, Qlik Sense proved to be a better choice.
What other advice do I have?
The data pipeline feature has helped my data processes significantly because whatever customers we are working with have their data in their own environment, whether on-prem or in their own cloud. Some of our customers have data in Amazon S3 buckets, while others have data in SQL servers. To extract the data from these sources and load it into the Microsoft Fabric environment, we need to build a pipeline. This is essential because until we create a pipeline and load the data from the client location to our location, we will not be able to build any dashboard on top of it. That is the essential step.
I assess Microsoft Fabric's security features as quite effective in meeting my organization's requirements. It is easy; while building the dashboard, we can create different roles. You just need to maintain a simple user table with user IDs and email IDs. You link those tables to your data tables so that the Power BI semantic model knows which row to show to which user or role. For example, as a manager, I can see more data than my employee; I can create these roles. Once we publish that dashboard to Microsoft Fabric, we can link these roles to the available users' emails using their Active Directory accounts. Microsoft provides security at various levels, including workspace access, report-level security, and database and table-level security. We can define security at various places, depending on our requirements.
I have not evaluated the effectiveness of Microsoft Fabric's data integration capabilities yet. I am speaking from my experience as a developer and architect who has worked with Microsoft Fabric. So I do not have quantifiable metrics to provide; I can only share insights based on my practical experiences and instincts.
Regarding the pricing of Microsoft Fabric, I find it quite affordable. Though I have not analyzed it from a long-term perspective, for someone starting with Microsoft Fabric or Power BI, the licensing cost is minimal. The Power BI license costs hardly ten dollars a month per user. As for Microsoft Fabric, while I am not entirely certain, typically one single user does not solely use the Microsoft Fabric environment; it will generally be procured by the company on an organizational account. Comparatively, the licensing is good, and the structure of licensing that Microsoft provides is quite strong, which I believe other tools could take inspiration from. I would rate my overall experience with Microsoft Fabric as a five out of five.
