In an enterprise CMDB with Discovery, Service Graph Connectors, cloud sources, integrations, application-owner input, and manually governed data, how should Identification and Reconciliation be architected when two legitimate sources disagree and neither source is universally authoritative?
A simply source-precedence hierarchy assumes that authority is static. In practice, authority may depend on CI class, individual attribute, lifecycle state, operating environment, business ownership, or the age ofthe observation.
What patterns have proven effective for establishing attribute-level and context-sensitive authority while still preserving deterministic reconciliation?
How do you maintain sufficient provenance to explain not only what value is currently on a CI, but whey the value won, which competing values were rejected, and whether the reconciliation decision remains valid as the CI moves through its lifecycle?
