Glossary

What is Data Mesh?

Data mesh is a decentralized approach to data architecture that treats data as a product and gives domain teams ownership of the data they know best, instead of routing everything through one central team. It was introduced by Zhamak Dehghani and rests on four principles: domain-oriented ownership, data as a product, a self-serve data platform, and federated computational governance.

The aim is scale. In a large organization a single central data team becomes a bottleneck as sources and demands grow. Data mesh distributes that work to the domains while shared standards keep the pieces interoperable.

A mesh decides who owns and serves data, not what state that data is in when a model runs on it. Each domain’s data product still has to be usable, reproducible, and traceable for the AI that consumes it, which is a separate readiness question the architecture alone does not answer.

Frequently asked questions

What problem does data mesh solve?

It removes the central data team as a bottleneck. As an organization's data sources and demands grow, one central team cannot keep up, so data mesh distributes ownership to the domains that understand each dataset while shared standards keep them interoperable.

What are the four principles of data mesh?

Domain-oriented ownership, data as a product, a self-serve data platform, and federated computational governance. Together they let domains own and serve their own data without a central bottleneck.

Is data mesh the same as a data lake or warehouse?

No. A data lake or warehouse is a storage and processing choice; data mesh is an organizational and ownership model that can sit on top of either. It changes who is responsible for data, not where it is stored.