Data Mesh Architecture

Data Mesh architecture embraces the concept of domain-driven design. It gives responsibility and ownership to the business domains who are most familiar with it, usually this is the domain that produces the data. The domain teams have access to tools to publish data and control the governance to share the data and metadata. Similar to domain driven development, contracts are used to control the interfaces between producers and consumers of data. The focus is on decentralising data hosting, making it easily consumable by different users across the organisation in a standardised manner

Data Mesh treats data as a product and applies product management and domain driven design principles to data. Data products can have SLAs around various aspects, including data quality, refresh rates, and even user satisfaction. By moving accountability from a centralised data engineering to the domain team that is closer to the data, the quality and trust of the data increases.

Data Mesh architecture enables domain autonomy by providing self-service capabilities to business teams. Data engineers build the necessary infrastructure to empower domain experts without requiring them to have data engineering skills. This infrastructure supports the self-service provisioning of data products.

Federated data governance describes how the organisation can define central data governance, but still allow data producers to manage and control their data products locally. This is implemented using a data platform and / or tools that allow data producers to create, share and manage access to their data products. Each domain is responsible for specific decisions, such as data models, quality assurance and data governance (i.e. who can access their data), whilst also complying with organisation wide governance rules. The data platform also allows data consumers to use data from multiple domains, once they have been granted access to them.