The Paradox of the Data Mesh
Although the data mesh calls for decentralizing development and data, the only way it can succeed is if you centralize what counts: data platform, data standards, data governance, and a data store. (See figure 1.)
That means you need a strong centralized data team that manages these things. And, also, the data team is on call to develop things that business domains won’t: like enterprise, cross-functional solutions, master data, and local solutions where no domain team exists.
The key to success in anything in data is to federate by aligning central and distributed resources. That is the difference between success and failure. That is what will keep your data mesh from turning into a data mush. (See figure 2.)
“The only way to success with a data mesh is to centralize what counts—and that requires a central data team.”