I read Zhamak Dehghani's book and I'm trying to reread it in between, though the author sells an idea that's more difficult to implement than a lakehouse and the whole construct becomes too complex to handle. Organizations have challenges to handle normal project work and reports, and they want to build a “partitioned” network of data products that are supposed to solve an organization’s data quality, governance problems and quality assurance? I’m sorry, but I can’t buy the idea! From where do you get the time needed for designing the products, negotiating the contracts, acquiring the technical resources needed to develop the data products, respectively the time for business users to be involved in this, etc. Most projects in the area fail because of organizational issues and not because of technology!

I can agree with the need to partition the data and solutions based on business domains, respectively assure that each domain takes the ownership, though the requirements for data products are too harsh to be implemented and as far I know there’s still no framework that allows by design to address the requirements of building, managing and monitoring data products. Microservices can work for stable entities like playlists, artists, or shopping lists, though I can’t imagine them yet to be able to handle complex transformations across big data repositories or complex systems like the ERPs.

Many sections from the book are just a good exercise for the imagination. It will be more than challenging to make it scale and the data mesh is supposed to be a matter of scale. Many ideas won’t stand the contact with the reality, not to mention that references like monolith, silo or strategic inflection point make me smile. I can accept metaphors and enthusiasm, though a Data Analytics infrastructure is not built on them!