Data mesh

Data mesh is an organizational and architectural approach in which responsibility for data is distributed to the domain teams that generate it, marketing, finance, logistics, rather than centralized in one data engineering team. Each domain publishes its own data product, with an owner accountable for its quality, documentation and contract, instead of handing raw data off to a central team to clean up and model.

This contrasts with a traditional centralized warehouse model, where a single data platform team ingests, transforms and serves data for the whole organization; that model concentrates expertise but often becomes a bottleneck as an organization grows. Data mesh also differs from data fabric, a more technology-centric approach that focuses on integrating and virtualizing access across existing data sources rather than on decentralizing organizational ownership.

Data mesh matters for organizations where a central team can't keep up with demand from many domains, each with different data and different priorities. It requires strong federated data governance, common standards for discoverability, quality and security enforced consistently across domains, or the result is inconsistent, hard-to-trust data scattered across teams rather than one bottlenecked pipeline. Because it is as much an organizational change as a technical one, many attempted adoptions fail on governance and incentives rather than on tooling.