Patterns / Data Architecture
Data Fabric
Metadata-driven integration that connects distributed data without forcing physical consolidation.
Sources
Active Metadata
Virtualisation
Policy
Consumers
Conceptual flow
Problem
Data is spread across estates that cannot be centralised for legal, technical, or cost reasons.
Context
Hybrid estates with strong metadata and lineage requirements.
Solution
Build an active metadata layer over distributed sources; use virtualisation, lineage, and policy enforcement to serve consumers without wholesale copying.
Benefits
- No mass duplication
- Central policy enforcement
- Faster discovery
Risks
- Query performance on virtualised sources
- Metadata quality dependency
Trade-offs
- Query latency against consolidation cost
When to use
- Regulatory data residency constraints
- Highly distributed estates
When not to use
- Small, already-consolidated estates
Related patterns
Discuss this pattern
How have you applied it — and where did it break? Bring your experience to the community.
Open the discussion