A CMDB is not merely a repository of infrastructure records. It is a representation of the reality. A map of where things are. The purpose of a CMDB is not to store data, but representation. Relationships, dependencies, ownership, and operational context.
Duplication breaks this concept.
When duplicates exist, the CMDB begins to fragment its understanding. Relationships attach to one CI but not the other. Incidents reference records to two instances of the same CI. Change histories become incomplete, one change related to one CI and another change mean to be related to this CI but It was related to the other CI that was found in the search, parallel realities. Despite this, things continue to function, but consequences tend to accumulate.
Duplicate CIs reveal the presence of multiple authorities attempting to describe the same entity.
Discovery tools, integrations, and manual entries. All of them, without clear identity boundary and reconciliation structures. It is like a system that accepts all.
Jesús Melendez 