What is Data Mesh?

Data Mesh is a decentralized data architecture paradigm proposed by Zhamak Dehghani in 2019 that transfers data responsibility from a central data team to business domain teams. Instead of a single data warehouse or data lake managed by a centralized team, each domain (marketing, finance, product) is responsible for publishing its data as "data products" with guaranteed quality, documentation, and SLAs.

Data Mesh rests on four principles: domain-oriented data ownership, data as a product, self-serve data platform infrastructure, and federated computational governance. These principles seek to resolve the bottlenecks of centralized architectures, where a single data team cannot scale to the pace of the organization.

It is important to understand that Data Mesh is not a technology or a product: it is an organizational and architectural model. It can be implemented with different technology stacks, but it requires significant changes in team structure, roles, and processes.

Centralized data architectures (monolithic data warehouse, data lake managed by a central team) face an organizational scalability problem. As the company grows, the central data team becomes a bottleneck: every new data requirement goes through the same queue. According to industry surveys, centralized data teams typically have 3-6 month backlogs.

Data Mesh proposes to solve this by distributing responsibility. The teams that know the data best (the domain teams) are the ones that publish and maintain it. The platform team provides the self-serve infrastructure. Governance is implemented in a federated way through computational policies (automated), not through centralized manual approvals.

In practice, adopting Data Mesh involves several changes. Each business domain identifies the data it produces and publishes it as data products with documentation, quality SLAs, and standardized APIs. The platform team builds self-serve tools so that domains can publish, discover, and consume data without depending on a central team.

Federated governance is implemented through automated policies: instead of a committee approving each access request, quality, security, and compliance policies are codified and executed automatically. For example, "every data product must have at least one active quality check" or "no PII field may be published without a classification."

Linedat facilitates Data Mesh implementation by providing the federated governance layer the model requires: domain-specific data catalogs, a shared glossary across domains, automated quality and compliance policies, and Data Steward roles assignable by domain so that each team manages its own data assets.