One of the four pillars of data mesh, as defined by Zhamak, is a distributed, self-service architecture. The other three pillars are not about technology. Is one out of four bad? 😀
So, the likely reasons people think data mesh is an architecture or technology are:
1) the first article by Zhamak says to move beyond the data lake so people assume it is a new technology to replace data lake;
2) because anything with hype in the data space seems to have a technology too;
3) someone saying “we are doing everything wrong re analytics” is such a bold statement, people can’t grok it immediately;
4) companies are trying to sell “data mesh technologies” and are muddying the waters; and/or
5) America-based engineers, as a group, want to focus on the technology so a paradigm focused on people and process is a foreign concept.
To us, the most crucial part of data mesh is the mindset change of data as an asset -> data as a product. The move to a distributed architecture is a logical next step after deciding to start with data product thinking.
If you have all the other pieces of data mesh in place but not data products and data product thinking, your data mesh implementation 1) isn’t a data mesh and 2) will be a disaster.