Data Fabric and Metadata-Driven Integration

What it is

A vendor describes it as a design approach, not a piece of software, that creates a unified view of data across environments. Organisations neither move distributed data to a single location nor take a completely decentralised approach.

The mechanisms usually named are data virtualisation, which makes data accessible without physically moving it, and active metadata, which is analysed continuously to automate integration and discovery.

There is no standard definition. The term is used mainly by vendors and analysts, and what a product called a data fabric does varies.

Why it matters

It is offered as an answer to data spread across many systems. It is often set beside data mesh: a fabric centralises through technology, a mesh decentralises through ownership.

DII has read only a vendor's description. It has read no independent evidence of adoption or of results, and holds its view of this subject with low confidence.

The problem it addresses

DII editorial view

Data sits in many platforms, and copying it all into one place is slow and sometimes impractical. A data fabric aims to connect it in place through metadata and virtualisation. How well that works depends on the product and on the state of the metadata.

Business and organisational relevance

DII editorial view

A data fabric is a way of buying and arranging technology. It centralises through tooling, where a data mesh decentralises through ownership, and the choice between them is about how the organisation wants to work.

Practical use cases

• A single catalogue and access layer over several platforms
• Querying data in place where copying it is impractical

Benefits and opportunities

• Less copying of data
• Consistent discovery and access

Risks, limitations and dependencies

• The term is defined mainly by vendors and analysts, and means different things in different products
• It depends on metadata quality, which is the hard part
• Virtualised queries can perform poorly at scale

Prerequisites for adoption

DII editorial view

• Metadata that is complete and current
• A clear definition of what the term means in the product considered
• Realistic query workloads to test against

Implementation considerations

• Ask what the term means in the product in front of you. There is no standard definition to hold a supplier to.
• Check the state of your metadata first. The approach depends on metadata that is complete and current.
• Test performance on real queries. Querying data in place can be slow at scale.

The DII Radar assessment

Catalogue status

Not yet assessed. DII has not published a Radar assessment for Data Fabric and Metadata-Driven Integration. It is listed in the catalogue so that it can be explained. No relevance, direction, strategic impact or enterprise adoption is given for it, and it is not plotted on the Radar.

What to consider next

DII editorial view

• Ask what the term means in the product in front of you
• Check the state of your metadata first
• Test performance on real queries