AI glossary · Advanced · Data engineering
Data mesh
Data mesh is an organizational and operating approach in which, instead of a single central data team managing all company data, each business unit (domain) that produces data takes responsibility for its own data and offers it like a 'product'.
About this term
- Level
- Advanced · Specialist terms that need the inner workings
- Category
- Data, ops & platforms
- Fields
- Data engineering
- Added
- 2026-10-09
How it works
Data mesh is commonly explained with four principles. First, teams for each business area (domain), such as orders, payments and logistics, take direct responsibility for their own data. Second, they offer that data as a 'data product' with documentation and quality standards so other teams can easily find, trust and use it. Third, a shared self-service data platform is provided so each team does not have to build infrastructure from scratch every time. Fourth, there is federated governance, where rules such as security and standards are set company-wide but carried out by each team.
Centralized approach vs. data mesh
Centralized approachCentral data team handles everything
- All requests pile up on one team
- Hard to deeply understand business context
- Easy to keep standards consistent
Data meshDomain teams take direct responsibility
- Data offered as a 'product'
- Uses a shared self-service platform
- Common rules through federated governance
공통 Both are operating approaches for making company data trustworthy and usable
Why it matters
As a company grows, all data requests pile up on the central data team, creating a bottleneck. The central team also finds it hard to understand the context of each business deeply and sometimes misinterprets what the data means. Data mesh is an attempt to reduce these problems by having the team that knows the data best take direct responsibility for it. A distinctive feature is that it is a concept about organizational structure and division of responsibility, not the name of a technical tool.
Things to know
Each domain team needs data skills and staff, so it can actually be a burden for small organizations. Decentralizing without common standards risks creating data silos where each team's data has a different format. That is why mechanisms such as data contracts, data catalogs and a shared platform are needed alongside it. In practice, many companies mix the centralized approach and the mesh approach.
Example
When a company with several business units adopts data mesh, the payments team, for example, builds a 'payment history' data product itself, registers it in the data catalog, and writes down the owner, update frequency and quality standards with it. The marketing team finds this data product in the catalog and uses it for analysis right away, without making a request to the central team.