Gartner Summit Recap 3: The Evolution of Data Architecture—from Fabric to Mesh and Beyond

Welcome to Part 3 of our Gartner Data & Analytics Summit recap series. If you’ve been following along, you’ve seen how metadata readiness and integration strategy are being redefined in the age of AI and data products. But underneath those conversations lie a deeper shift in how organizations structure their entire data architecture.

These three sessions in particular stood out:

Henry Cook, VP Analyst at Gartner, shares his thoughts on how data fabric has evolved into a foundational architecture for simplifying access, governance, and optimization across increasingly complex data environments.

Michele Launi, Senior Principal Analyst at Gartner, explores how data mesh can help to decentralize data management—not just architecture—to enable autonomy, scalability, and agility. He outlines the four pillars and shares six best practices for successful implementation.

Ehtisham Zaidi, VP analyst, and Robert Thanaraj, Senior Director Analyst, posit that it’s time to end the data mesh versus fabric debate by showing how these two frameworks are better leveraged together to meet a variety of goals.

Combined, these sessions made a compelling case that data fabric and data mesh are not competing strategies—they’re complementary frameworks that, when used in tandem, support scalable, domain-driven data delivery.

Fabric, explained

In his session on data fabric, Henry Cook offered a refreshingly pragmatic take: Data fabric isn’t a shiny new platform; it’s an architectural layer that connects everything you already have. From warehouses and lakes to virtualized and federated systems, data fabric ties it all together using active metadata and automation to reduce friction across the data ecosystem.

Henry framed data fabric as a response to modern complexity. With data scattered across clouds, applications, and business units, organizations need a way to unify access, optimize performance, and surface insights in context. That’s where active metadata comes in—enabling the system itself to recommend improvements, trace lineage, anticipate cost, and even adjust pipelines dynamically.

Henry’s view: Data fabric isn’t a rip-and-replace strategy for existing architectures. He emphasized that the goal is not to rebuild your architecture, but to enrich it with context, observability, and intelligence. The result is a more adaptive foundation for data products, AI initiatives, and agile development at scale.

Mesh, explained

Michele Launi’s session on data mesh reinforced the model as an organizational and operational shift rather than a technical implementation. Mesh reframes data strategy through decentralized ownership, giving domain teams control and responsibility for data treated as a product.

Michele emphasized the core principles of data mesh as practical foundations for scaling decentralized data management:

• Domain ownership
• Data as a product
• Self-service infrastructure
• Federated governance

Together, these enable business units to take responsibility for the data they generate—curating it, managing its lifecycle, and ensuring it delivers value. But Michele was clear: Success with mesh requires structural change. Rather than launching full-scale mesh initiatives, he encouraged starting small—identifying promising domains, building cross-functional fusion teams, and applying a product management mindset.

Mesh is a cultural and organizational shift. It works best when approached iteratively, with a clear focus on value delivery and strong collaboration between business and technical teams.

Challenging the fabric versus mesh debate

In a session consisting of equal parts insight and improv, Ehtisham Zaidi and Robert Thanaraj made a compelling case to stop treating data mesh and data fabric as opposing choices. According to them, it’s a category error: Fabric is an architectural design, while mesh is an operating model. They leaned into a familiar metaphor: Fabric and mesh are often treated like apples and oranges—fundamentally different and hard to compare. But as Ehtisham put it, they’re also incomplete without each other.

Fabric provides the technical foundation—connecting systems, analyzing metadata, and enabling intelligent automation across the ecosystem. Mesh, on the other hand, distributes responsibility, helping domain teams manage and deliver data products that are aligned with business needs.

Gartner’s research supports this combined approach. Only 13% of surveyed organizations are implementing fabric and mesh together, but those that are show higher success rates with data product delivery. Unsurprisingly, the biggest blocker isn’t strategy—it’s metadata maturity. Without active, actionable metadata, neither model can reach its full potential.

The bottom line: Fabric and mesh aren’t rivals. They’re partners. “Apples need oranges,” Ehtisham quipped, underscoring that a modern data strategy depends on both structure and decentralization working together.

Final thoughts

Data mesh and data fabric aren’t mutually exclusive—they’re two sides of the same coin, each addressing different challenges on the path to modern data architecture. Fabric provides the technical backbone, enabling connectivity, automation, and observability across a fragmented data ecosystem. Mesh builds on that foundation by shifting how data is owned, delivered, and governed—bringing responsibility closer to the business.

Think of it this way: Fabric makes your systems smarter. Mesh makes your teams stronger. Together, they support a data strategy that’s both technically resilient and operationally adaptable—something each session echoed in its own way. Whether it’s improving integration maturity, enabling decentralized delivery, or building a more responsive architecture, the path forward isn’t one or the other. It’s both.