While the Data Fabric tolerates some amount of distributed data governance and management, it ultimately moves an enterprise in the direction of centralized control over data in the Fabric in order to maximize the reliability of the metadata and output of the machine learning algorithms. In contrast, the Data Mesh more completely embraces distributed data governance and management. Different teams of domain experts maintain control over their own datasets and make them available directly to other teams via “data products” without any requirement to go through a centralized process. There is no built-in mechanism to discover new datasets relevant to a particular analytical task, but a global catalog can be used to facilitate data discovery.