Why Storing the Same Data in Two Places Will Eventually Break Your System

Software engineer Serguey Asael Shinder argues that maintaining the same piece of data in multiple locations within a system is a reliability risk that will inevitably lead to inconsistencies. He explains that data drift does not require a bug to occur — a single out-of-order write is enough to cause two values to diverge permanently. Shinder warns that silent auto-repair jobs are particularly dangerous, as they mask the underlying problem rather than surfacing it. He acknowledges that copying data across services is sometimes unavoidable, but insists that such duplication must be tracked, reconciled openly, and never treated as a true source of truth. His core principle: every fact should be stored in exactly one place, with all other references treated explicitly as derived copies.