What is a data fabric?
A data fabric is an architectural layer that spans a company’s databases, applications, cloud storage, and on-premises systems and presents them as one connected, governed set of data. Instead of copying everything into a single repository first, the fabric catalogs what exists where, applies common security and quality rules, and lets users and applications query or move data through one interface. Active metadata does much of the work: the fabric tracks schemas, lineage, usage, and sensitivity across sources and uses that knowledge to automate tasks such as flagging a table that holds personal data. In practice a fabric is assembled from a catalog, integration tools, a query layer that reaches into multiple sources, and policy enforcement.
Why a data fabric matters
A regional bank holding company with three core banking systems and two loan origination platforms cannot migrate them all before it needs a consolidated view of a borrower. A fabric gives analysts and AI agents one governed path to all five while consolidation proceeds at its own pace. Compared with a data mesh, which changes who owns the data, a fabric changes how the data is reached and leaves ownership where it is.
What to watch for
- Federated queries against live source systems can slow those systems, so confirm which workloads will run against copies.
- Vendors use the term loosely, so ask which capabilities (catalog, lineage, policy enforcement, virtualization) are actually included.
- Metadata quality decides whether the automation helps or misleads, and someone must own it.