Combining Central and Decentral Data Platforms
A short story about costs, complexity and responsibility
Central data platforms can bring great efficiency and positive cost effects while decentralized or departmental data platforms are near to the business effectivity.
How it started…
At a customer with a very decentral history and high autonomy of business departments, we are establishing a central data platform with a federated responsibility and business ownership of data products. While the central platform offers a high degree of autonomy, decentral data platforms can make sense, especially if they are strongly bound to business logic. Here a Customer Data Platform for sales and marketing or Jedox as a planning and consolidation solution are data intense but can manage business logic a general purpose data platform can’t.
How it is going…
There is no final or perfect solution to balance costs and complexity, so this is a work in progress. The first priority is to establish a clear ownership and creating synergy effects. Decentralization will increases the overall cost which could be worth it.
Outlook
A Headless approach distributing data via a data products can be very efficient, while a Data Fabric approach best reduces the complexity of data access in decentral environments while better supporting centralization. A Data Mesh-like multi-platform approach needs a high maturity, decentralized responsibilities and a data catalog/data marketplace or possibly a data product management solution to support this. For a mature company with a data-driven approach, this can make a lot of sense.