We believe this misconception comes from trying to sell data fabric products as a data mesh. The term data fabric is hard to define but vendors have jumped on it pretty hard. Those same vendors (and some new ones) are trying to do the same to data mesh…
To be clear, data mesh is much more than a technology or architecture.
There are a number of companies saying they sell a data mesh as of around February or March 2021. Some are large enterprise companies where you might expect some marketing overreach – HP Enterprise, KPMG+IBM, Oracle have all transgressed here. There are also a number of emerging startups that are trying to do the same but we will refrain from calling them out just yet.
The three big cloud vendors are at least toeing the line a bit better; but even AWS/Azure/GCP are saying you can build a data mesh with their cloud.
Given that data mesh is much more than an architectural or technology-based approach, saying you “offer a data mesh” is, quite frankly, stupid. Phrasing it as “we offer a self-service data platform for your data mesh” would make far more sense and lose less credibility.
To be clear, technology/architecture is a crucial part of data mesh. But none of these companies are including “data product thinking” as part of their offerings and thus, are not a data mesh as defined by Zhamak 😀
We are on the lookout for vendors that are doing it the right way and will update this page as they emerge. Thus far, Snowflake, Databricks, and (mostly) Starburst have not overreached on this and we thank them. GCP is probably the best behaved of the big three cloud vendors and Azure isn’t too far behind.