Event Sourcing Trade-Offs: Weighing Production Costs Against Its Theoretical Benefits

The article argues that Event Sourcing looks ideal in theory, offering an immutable, timestamped event log with full auditability and the ability to reconstruct past states. In production, however, the append-only log accumulates permanent storage weight and makes schema evolution difficult, since historical events cannot be rewritten and must instead pass through upcasters. Operational costs include upcaster chain bloat that burdens CPU and memory, projection lag causing stale reads under heavy write loads, and harder debugging across distributed event chains. The author suggests Event Sourcing suits domains needing strict, legally binding audit trails with stable events, while unsettled domains or teams lacking operational maturity should default to a conventional state-backed model.