Decentralized Data

Data mesh became too popular too quickly. I was listening to the recent ThoughtWorks podcast on revisiting Data Mesh.

The essence of data mesh is decentralized management and value generation using data.

The essence of data mesh is decentralized management and value generation using data. This is an important differentiation. It affects how we describe data using analogies.

One of the commonly used analogies is that "data is the new oil". This might help to describe large AI-driven initiatives, where large-scale data is mined, sourced, transported, stored and consumed constantly to build AI engines capable of producing general intelligence.

In a typical enterprise, data may not be the new oil. If anything, data could be the currency, a medium of exchange to consolidate or transfer value from one function to another. Currency is deposited for the purpose of liquidity, for immediate needs. Otherwise, it is converted into assets or exchanged for goods and services. Thus, data products and data services are the assets built using data. Even the narrow AI models required in the context of a specific business are just data products.

So, consider how you generate value from your data? Are you just hoarding data or converting into assets? Are there necessary and sufficient data stored to rebuild your assets when it is depreciated, expired or destroyed? Are you generating value locally within business functions and exporting it globally?