CMDB and CSDM Data Governance, Health, and Maintenance
Governance makes the CMDB trustworthy
A CMDB degrades without governance. Candidates must know the mechanisms that keep data healthy: clear ownership of each CI class and attribute, data standards (naming, mandatory fields, allowed values), change control over the model itself, and audit cycles that compare recorded to actual state.
Concepts to name
- Data stewardship: named owners accountable for accuracy of a class or attribute.
- System of record and system of entry: where truth lives versus where it is edited.
- Health metrics: completeness, correctness, compliance with standards, duplication rate, orphan records, and relationship coverage.
- Certification and attestation: periodic confirmation by owners that records are still valid.
- Lifecycle and retirement: retiring CIs and services so stale records do not distort impact analysis.
- Change management integration: model changes treated like any other change.
Traps
A frequent trap is equating health with record count - a large CMDB can be mostly stale. Another is assuming automated discovery solves governance; discovery without ownership and reconciliation produces duplicates and drift. Questions also test that governance applies to the model (classes, relationships) as well as the data.
Worked example
A quarterly health review finds 12% of server CIs missing an owner. The steward assigns ownership rules by support group, adds owner as a mandatory field, and schedules monthly reconciliation against discovery. Completeness is tracked as a trend, not a one-off cleanup.
Rule of thumb
Measure health by fitness for the decisions the CMDB supports. If no one consumes a metric, do not chase it.