The situation
The client’s CMDB covered more than 40,000 configuration items across on-prem infrastructure and AWS. A previous implementation partner had populated it using a broad discovery scope. Within 18 months, the data quality score had dropped to 34%, below the threshold the change management team needed to rely on it.
Internal audit had flagged the CMDB as a control weakness. The platform team needed to remediate it before the next audit cycle.
What we did
We started with a two-week CMDB health assessment to identify why the data quality had deteriorated. It found three problems:
- Discovery ran on a 72-hour cycle, leaving ephemeral cloud workloads with stale data
- Deduplication rules did not account for cloud-specific CI naming conventions
- Service relationships came from a one-time import that had never been reconciled
The rebuild focused on discovery schedules, record matching, and the cloud data feeds. We rebuilt the discovery schedule and rewrote deduplication rules for AWS EC2 and RDS naming patterns.
We also implemented Service Graph Connectors for the client’s three primary cloud accounts.
Results
The CMDB data quality score increased from 34% to 91% over the 14-week engagement.
- AWS CI staleness fell from 72+ hours to under 2 hours
- The change management team reinstated the CMDB as the authoritative source for impact analysis
- The internal audit finding was closed at the next review cycle