Configuration Management Database
About 15% of the CIS-Discovery exam — see all 4 domains.
Discovery exists to populate a healthy CMDB. This domain is about how data lands correctly and stays trustworthy.
CI classes and the CMDB
- CIs live in class tables that extend a base class (
cmdb_ci) in a hierarchy (e.g.,cmdb_ci_server→cmdb_ci_win_server). A more specific subclass inherits the parent's fields. - Classification during Discovery determines which class a CI is written to; reclassification moves a CI from one class to another (for example, when a device was first found generically and later identified more precisely). Reclassification has three types: upgrade, downgrade and switch. It is controlled by the glide.class.* system properties and by reclassification restriction rules, not by a setting on the identification rule.
Identification and Reconciliation Engine (IRE)
The IRE is the gateway through which Discovery and other IRE callers write CIs, so the CMDB stays free of duplicates.
- Identification rules: each CI class has an identification rule made of one or more identifier entries, evaluated in priority order. An entry lists the criterion attributes that must match an existing CI for it to be considered "the same." Example: a server may be identified by serial number first, then by name + IP.
- Reconciliation rules: when multiple discovery sources can write the same attribute, reconciliation decides which source is authoritative for that attribute. Once an authorized source has updated an attribute, later updates are accepted only from the same or a higher-priority source (unless the data is stale under a data refresh rule).
- Duplicates: by default (skip_duplicates true) when an identifier matches several duplicate CIs, the IRE updates the oldest, marks the others with duplicate_of, and creates a de-duplication task.
- Data sources & precedence: IRE tracks which source last updated a CI; this underpins reconciliation and helps spot conflicting writes.
- IRE applies to payloads from its callers (Discovery, IntegrationHub ETL, Service Graph Connectors and the IRE APIs), inserting, updating, or matching CIs as the rules dictate. A plain import that writes CMDB tables directly does not pass through the IRE.
CMDB Health
CMDB Health dashboards score data quality across KPIs grouped into three families you should memorize:
- Completeness: are required and recommended fields populated? (e.g., CIs missing an owner or a critical attribute).
- Correctness: are CIs free of duplicates, orphans (per-class orphan rules that can test attributes, missing relationships or specific relationship types) and stale records (not updated within an expected window)?
- Compliance: do CIs pass the audits (certification filters and templates) defined for them?
Health jobs run on a schedule and feed dashboards used to prioritize remediation (deduplicate, fill gaps, retire stale CIs).
What to drill for Domain 3: the CI class hierarchy and reclassification; that IRE is the entry point Discovery uses to prevent duplicates; identifier entries with criterion attributes evaluated by priority; reconciliation deciding authoritative source per attribute; and the three CMDB Health families (completeness, correctness, compliance) plus the fact that duplicate, orphan and stale CIs are Correctness metrics.
Sample questions from this domain
Three of the 43 in this domain, with the reasoning. The full set is in the question bank.
What is the primary purpose of the Identification and Reconciliation Engine (IRE) in the CMDB?
- A. To identify whether incoming CI data matches an existing CI and to reconcile attribute updates from multiple sources ✓
- B. To classify each device by the port probe that answered and then launch the matching identification and exploration probes
- C. To record which credential and MID Server last succeeded for each host so that later Discovery runs reuse them first
- D. To calculate Completeness, Correctness and Compliance scores for each CI class shown on the CMDB Health dashboard
Why: The IRE decides whether incoming CI data matches an existing CI (identification) and which source may update which attributes (reconciliation). Classifying devices by the answering port probe and launching later probes is the Discovery classification phase, not the IRE. Remembering the credential that worked for a host is credential affinity, a Discovery behavior. Completeness, Correctness and Compliance scores are produced by CMDB Health jobs, which measure data quality rather than matching or reconciling payloads.
Which CMDB component defines the attributes used to uniquely identify a CI of a given class?
- A. Reconciliation rule with a datasource priority per attribute
- B. CI Identifier (identification rule) with identifier entries ✓
- C. Discovery classifier with its classification criteria
- D. Data refresh rule with an effective duration for the class
Why: A CI Identifier (identification rule) holds the identifier entries, each listing the criterion attributes the IRE uses to match a CI of that class uniquely. A reconciliation rule decides which data source may update which attributes after a match, not how the CI is found. A Discovery classifier decides which class and probes apply to a device during classification. A data refresh rule sets how long a source's data stays current before a lower-priority source may update it.
What is the role of 'reconciliation rules' (datasource precedence) in the CMDB?
- A. They define which identifier entries the IRE evaluates first when it matches an incoming payload to existing CIs
- B. They merge duplicate CIs automatically by keeping the record most recently updated by any discovery source
- C. They decide which CI class a payload is moved into when two sources report different classes for one device
- D. They control which data source is authoritative for specific attributes, preventing lower-priority sources from overwriting them ✓
Why: Reconciliation rules set datasource precedence per attribute, so once an authoritative source updates an attribute, a lower-priority source cannot overwrite it (exceptions: it is the first source to update the CI, or the CI is stale under a data refresh rule). The order in which identifier entries are tried is the priority set on the identification rule. Duplicates are handled through de-duplication tasks, and there is no most-recent-writer rule. Class changes are reclassification, governed by glide.class properties and reclassification restriction rules.