The middle ground between canonical models and data mesh

Canonical Data Models create God Objects that grow uncontrollably as every department adds fields, while pure Data Mesh introduces complexity with multiple anti-corruption layers. A hybrid "Federated Hub-and-Spoke" approach combines both: maintain a lean central CustomerIdentity table in a data lake, extend it with domain-specific tables (SalesCustomer, SupportCustomer), use events to sync base data centrally while keeping domain data in source systems, and send only base customer data between services. This preserves bounded contexts, supports legacy systems, enables analytics, and avoids massive payloads—though it requires discipline to keep the base table lean and careful handling of race conditions.