Shifting Pillars: The refinement of Microsoft Fabric’s core principles

Author: Briton Wells
Solution Architect, Data & Dynamics
Long View

 

---

 

Lots of announcements this week from the European Microsoft Fabric Community Conference!

Definitely take the time to read through Arun’s blog here.

I won’t rehash the feature announcements themselves. Rather, I wanted to focus on the changes to the “pillars” of Microsoft Fabric and consider what these changes represent for Fabric going forward.

In May 2023 when Fabric was announced, the four core principles were:

1. Fabric is a complete platform
2. Fabric is lake-centric and open
3. Fabric can empower every business user
4. Fabric is AI Powered

Today, the core pillars have been reimagined to:

1. AI-powered development
2. Lake-centric and open (unchanged)
3. AI-enabled insights
4. Enterprise ready

All of these are interesting in their own way – there is a mix of unchanged, removed, and totally new.

Removed

The main pillar being removed is the concept of Fabric as a complete platform. I do think this is absolutely still part of the foundation of Fabric, being an integrated SaaS offering, however other products are evolving into complete platforms as well – this won’t be a major differentiator long-term.

Unchanged

Unchanged is the core concept of Fabric being lake-centric and open. This certainly is one of the fundamental advantages of Fabric over other products. Shortcuts and Mirroring continue to evolve, the data ingestion tools are maturing, and OneLake continues to resonate with customers.

Refined

The AI pillar – “Fabric is AI Powered” has been more refined than changed dramatically. I think this is a good change – Fabric itself was not “powered” by AI, rather it empowered both developers and users to build their data estate and generate insights from that data. The two new pillars reflect this split – AI-powered development, and AI-enabled insights. We’re now thinking through each AI feature as either for the developer, or for the end user of the data. More clarity here is good.

New

The new pillar, and I think the most interesting change, is the brand new pillar – Enterprise ready. Let’s dig into that a bit. From Arun’s blog “You can confidently deploy and manage Microsoft Fabric with category-leading performance, instant scalability, shared resilience, and built-in security, governance, and compliance.”.

I’m happy to see the reassurance here – Fabric is being positioned as a serious contender in the enterprise analytics market. On the flip side, there is still a fair amount of fundamental functionality in preview, still on the roadmap (not in public preview yet), or not even on the current published roadmap.

The changes being announced today under this pillar relate to:

• Fabric Runtime 1.3 GA
• Trusted workspace access and Managed Private endpoints in all fabric capacities (Previously announced) and now available in Trial capacities (new)
• Deeper integration with Microsoft Purview
  • Information protection labels can control access to Fabric items
  • DLP Policies against Fabric Lakehouses
• Fabric item tagging and domain enhancements

These are all amazing announcements – there was some clear feedback on “gating” network security behind an F64 capacity. It meant that even small development workloads either could not leverage these features (often a showstopper) or that all workloads had to be on F64 and above (quite restrictive in terms of the overall architecture). Now we are free to right-size capacities without major feature differences (well, other than Copilot!).

Our customers LOVE Microsoft Purview. Closely aligning with the updates on the Fabric side, it would be remiss to not mention that Microsoft Purview Data Governance has just gone GA. We’re seeing renewed interest in both Purview itself as well as how Purview and Fabric can provide data governance at scale.

Where I think organizations will continue to see some gaps:

• The overall process, procedures, and support of Git integration and deployment pipelines (Dataflows not supported, difficult to parameterize all Fabric items to work well between workspaces)
• Missing connectors that require workarounds such as invoking a pipeline in Synapse or Azure Data Factory (however, it’s much easier to invoke said pipelines as of today!)
• Lack of a unified security model – this is still on the roadmap, but the discussion today is confusing with customers, as security can be defined in many different areas. This is likely the biggest remaining challenge with enterprise deployments today

Subscribe to our newsletter for the latest updates.