A common pain point organizations face in their data architecture is the interfacing process between systems, including managing multiple different tools to create a coherent, functioning environment. Microsoft listened to its users’ pain points and delivered a one-stop-shop solution in the form of an end-to-end cloud-based software-as-a-service (SaaS) called Microsoft Fabric. It features tools for many data and analytics needs by including:
- Storage via OneLake,
- Data Integration by Data Factory,
- Data Engineering, Warehousing, Science, and Real-time analytics in Synapse,
- Business Intelligence with Power BI,
- Action platform via Data Activator, and
- Governance within Purview.
All of these tools come together to produce an all-in-one analytics solution to meet data migration, data science, analytics, warehousing, and business intelligence needs. The overarching goal of this solution is the simplification of invoicing and a noisy data environment. While there are added benefits in a coherent end-to-end solution, there are also drawbacks to this design as well. Organizations should weigh both before making a change to their platform.
There are a few key advantages that Microsoft has woven into Fabric primarily benefiting smaller or start-up organizations, although they can help organizations at all maturities. These features are a great jumping off point, as they streamline the data pipeline and limit the moving parts to worry about.
A single invoice simplifies the billing process and may save an organization money. Rather than paying multiple vendors for their single tool, Microsoft Fabric would be a single vendor with a single invoice per period. This reduces overhead with contract approvals, as there’s only a single initial setup and renewals, as well as decreases accounting time since there’s only one account payable to manage.
Microsoft Fabric also offers a flexible billing plan with billing units to track the service usage of each tool within the solution. This provides capabilities for organizations to optimize usage of each tool to reduce cost as well as understand the direct lines between activities in one of the modules and the cost impact.
Having an all-in-one solution can cut interfacing struggles down tremendously, as the tools seamlessly have the data flow through rather than involving middleware and extensive data translations. This reduces integration time, security issues, and data discrepancies. When separate systems are present, integrations must be properly configured and connected to each tool stemming from the data warehouse. This multi-step data transport system can allow security leaks and mistranslations between each differing system.
Microsoft Fabric also avoids issues around synchronizing data formats and time zones within the system. These may seem like small details, but managing and monitoring platforms for these issues can be time-consuming and costly for lean data teams. With Fabric, these discrepancies can be minimized and managed through a single platform that can proactively avoid these issues.
Lastly, this all-in-one nature of Microsoft Fabric unlocks new automation capabilities. While no features have been announced in this space, an all-in-one platform creates the opportunity to trigger actions in one tool based on activities in another. For example, adding a new aggregated table or metric in the semantic layer could trigger a workflow that notifies an analyst to populate a data catalog entry, explaining those metrics. This unified tool could provide a Generative AI tool with the inputs required to identify possible testing gaps to reduce bad quality data.
A Delta Lake
Storage with OneLake adds a delta lake, improving the foundation to store enterprise data. Delta lakes further break down compute and storage by separating storage into data and the structure. This drives the cost of storage down and provides new capabilities for data teams.
For example, Delta Tables provide additional data ingestion capabilities, allowing both batch and streaming data and preventing the locking of objects during load. The structure also allows easy auditing of changes and revisions to structure, and the ability to “time travel” back to former versions of these tables. Beyond that, the Delta Lake provides optimization capabilities and the ability to evolve schemas to manage changing inbound data formats.
However, teams should be cautious with this new technology. Traditionally SQL or database using data teams may struggle with migrating to this source and the nuances of management. Seeking out new resources or an analytical partner can help manage this transition.
Next, we will dive into some further observations about the overall design of Microsoft Fabric and situations to consider when deciding if this would be the best fit for your business.
What It Doesn’t Do:
While Fabric does offer an all-inclusive software, most of the tools (for example, Power BI and Synapse) are already on the market, so Microsoft appears to just be repackaging some tools into a “new product”. Beyond that, some of these tools offered may not be considered best in their spaces, so while you have this all-in-one model, some parts may be subpar in a market comparison.
Another drawback to the comprehensive tool package is the possibility for a tool sprawl in a mature organization. A mature organization would have its data architecture much more cemented with its customizations in place. Mature organizations wouldn’t need all of these tools, so implementing an entirely new end-to-end system may make it more complicated. Instead, it would be more efficient for these types of organizations to focus only on the problem areas in their data structure. This is something Microsoft Fabric doesn’t offer and is probably its biggest setback, which is a lack of customization of the tools. This would also cause a problem in not being able to accommodate nuanced data engineering and architecture scenarios.
Is Microsoft Fabric Right for Your Organization?
While there’s a large amount of hype around this package, it may not be the category killer that it has set out to be. It’s important to keep in mind that this is still only the preview version, so more features may be revealed once the full release happens. However, based on the preview version, it can be concluded there are likely more disadvantages than advantages to implementing this solution for certain organizations. There are limited scenarios where this is potentially the best path, but it would have to clearly be a right fit situation.
If you would like to dive deeper into how this or other tools could fit into your data environment, reach out to our data experts, as they can help advise on what may be the best option for your business.
Subscribe to Clarkston's Insights
Contributions by Matt McMichael, Sarah Radebaugh