Concept card

BARC — The Data Fabric Survey

As of 2026-10-06

Edited by Cédric Mary, sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

What is BARC?

Data fabric is a technology pattern (active metadata, virtualisation, lineage); data mesh is an organisational model (domain-owned data products) — conflating them misdiagnoses which problem a client actually has.

Data fabric is an architectural pattern, not a product you can buy and switch on — a distinction that gets lost constantly in vendor marketing. It describes a design principle: that data integration, access, and governance across an enterprise should be driven by metadata, specifically active metadata, rather than by hand-coded point-to-point pipelines connecting every source system to every consuming application. The promise is a unified, queryable, governed view of enterprise data regardless of where it physically sits — on-premise ERP, cloud warehouses, SaaS applications, data lakes, or operational databases — without requiring every byte to be copied into one place first.

The three mechanisms

A working data fabric rests on three capabilities working together. Active metadata management goes beyond a passive catalog that simply records where data lives; it tracks relationships between datasets, observed usage patterns, schema-change history, and computed quality scores, and it uses that intelligence to recommend joins, flag drift, and automate policy enforcement rather than requiring a human to notice a problem. Intelligent data virtualization lets a consumer query across genuinely heterogeneous sources through a semantic layer without first physically moving all of that data into a central store — the query is federated out to where the data lives and the results are assembled on the fly. Automated lineage tracks the provenance of every data point from its origin through every transformation to its point of consumption, which is what makes impact analysis possible when a source schema changes upstream, and is frequently a regulatory requirement in its own right.

Why it matters

Active metadata goes beyond a passive catalog — it tracks relationships, drift, and quality scores to recommend joins and automate policy enforcement.
A data fabric can be implemented by one centralised engineering team; a data mesh requires distributing ownership across business domains like Finance and Supply Chain.
Lineage enables impact analysis when a source schema changes and supports regulatory data-tracing requirements.

Key points

BARC user-experience survey for the data-fabric category.
Covers data integration + metadata + catalog + quality + governance + orchestration.
Vendor field: Informatica, Talend / Qlik Talend, Microsoft Fabric, Denodo, etc. (BDC competes here).
Newer survey than BI / Planning — category itself is recent.
Rankings paywalled.
Data Fabric Survey 26 (BARC's 7th-year edition) sampled 776 participants across 45 countries and 19 products; 69% report high/very high benefit for data accessibility and 68% for data control and trust, but 36% see little or no benefit from current tools for AI readiness (BARC, 2026).
Vendors market AI/ML and data-product capabilities far more than users actually deploy them: 42% of vendors position AI/ML as a differentiator versus only 18% of users reporting active use, with a similar gap for data products (51% vs 26%) and data governance (41% vs 25%) (BARC, 2026).
In active tool-selection processes, Microsoft Fabric is the most frequently evaluated platform (26%), ahead of Databricks (21%) and Snowflake (19%) — the shortlist a consultant scoping SAP BDC should expect a client to already be building (BARC, 2026).

Terms used on this page

Data fabric: Integrated data-management architecture combining ingestion, metadata, catalog, quality, governance and active orchestration under unified semantics — Gartner-popularised term.
Reusable IP: An artifact, checklist, or model that can be reused across clients without copying client-specific data.
Data mesh: A sociotechnical operating model, distinct from data fabric, that distributes data ownership to domain teams (Finance publishes Finance data products, Supply Chain publishes its own) under federated governance rules each domain applies independently, rather than a central team owning all integration.
Active metadata: Metadata that goes beyond a passive catalog entry recording where data lives — it tracks relationships, usage patterns, schema-change history and quality scores, and uses that intelligence to recommend joins, flag drift, and drive automated policy enforcement rather than requiring a human to notice a problem.
Data virtualization / federation: Letting a consumer query across heterogeneous sources through a semantic layer without first physically moving all the data into a central store — the query is federated out to where the data lives and results are assembled on the fly; SAP Datasphere's remote tables are a concrete implementation.
Lineage (data lineage): Tracking the provenance of a data point from its origin through every transformation to its point of consumption; enables impact analysis when a source schema changes upstream and is frequently a regulatory requirement in its own right.

Sources

Full card available to members. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →