ServiceNow CIS-DF: CMDB Reconciliation Rules

Reconciliation rules answer a question that appears only after identification succeeds: when several trusted sources describe the same CI, which source is allowed to change each attribute? Without that policy, the CMDB can become a competition between integrations where the latest write wins, values oscillate between sources, and nobody knows which field to trust.

In ServiceNow CIS-DF data-foundation work, reconciliation defines predictable source authority for CI attributes. A cloud connector, Discovery, import process, manual owner, or another integration can all write to the same CI, so the platform must know which source is authoritative for each class and attribute.

Platform engineering depends on stable reconciliation because source authority affects automation, reporting, service impact, security workflows, and incident response. A field that changes unpredictably can break downstream decisions even when the CI itself is correctly identified.

Identification establishes the CI; reconciliation establishes authority

ServiceNow’s IRE separates identity from update authorization. Identification rules determine which existing CI the incoming payload matches. Reconciliation rules then determine whether that discovery source can update the attribute. Keeping those decisions separate prevents a common design mistake where teams alter identity logic just to stop one source from overwriting another.

The paired identification rules should therefore be stabilized first. If two sources are writing different CI records, reconciliation is not the fix. Once both feeds target the same CI, attribute authority can be designed without creating duplicate identity.

This separation becomes critical during troubleshooting. A field that refuses to change even though the correct CI was found points toward reconciliation or source authority. A brand-new duplicate record points toward identification. Knowing which phase failed prevents teams from modifying the wrong IRE control and creating a second problem.

Define authority at the attribute level where sources genuinely differ

A source may be authoritative for some attributes and weak for others. A cloud platform can reliably report resource ID, region, and native state; an endpoint system may know operating-system build and installed software; an application owner may control business criticality. Treating one source as globally authoritative can erase useful data from others.

Reconciliation should reflect those differences. Teams should group attributes with the same ownership logic, document why the source is trusted, and avoid microscopic policies that nobody can maintain. The objective is understandable authority, not the largest possible number of reconciliation definitions.

Attribute groupings should follow business meaning. Technical lifecycle fields, location, ownership, software versions, and business criticality often come from different systems. Designing a single source priority for all of them can make the CMDB internally consistent but factually wrong because no one source is authoritative for every dimension of the CI.

A practical way to discover attribute ownership is to trace each field back to the business process that produces it. Technical attributes may originate from discovery, finance attributes from asset systems, and service ownership from application governance. When authority follows the actual process that creates reliable data, reconciliation becomes easier to defend than a simple “source A outranks source B” hierarchy.

Static and dynamic rules serve different governance needs

ServiceNow supports static reconciliation rules that define source priority for attributes and dynamic reconciliation behavior for more flexible source authorization. The design choice should follow how stable the source relationship is. A simple, durable source hierarchy may fit static rules; more conditional authority may require dynamic behavior.

Complexity has an operational cost. Dynamic rules are powerful, but a rule nobody can explain during an incident is a weak control. Teams should prefer the simplest policy that correctly represents authority and use test payloads to show which source wins under each meaningful condition.

Dynamic rules should have explicit test cases that show when the condition changes authority. If the rule depends on source state, recency, or another attribute, operators need to know why one payload was accepted and another was rejected. Otherwise the flexibility becomes invisible behavior that only platform specialists can diagnose.

Source identity must be consistent across connectors and imports

Reconciliation decisions depend on the discovery source recorded with incoming data. If the same integration writes under different source names across environments or versions, the platform may treat those writes as separate authorities. Source naming is therefore part of the interface contract.

For Service Graph Connectors, connector upgrades and custom integrations can introduce new source identities. Before changing a source value, engineers should audit existing reconciliation rules and verify that the new identity inherits the intended permissions.

Source-name governance also matters during migration. Replacing a legacy connector with a Service Graph Connector may cause the same real system to write under a new discovery source. Before cutover, teams should decide whether reconciliation definitions must be cloned, retired, or re-prioritized so authority does not disappear during the transition.

Data source rules prevent unauthorized CI creation

Reconciliation controls updates to existing CIs, but some sources should also be prevented from creating new records for particular classes. ServiceNow data source rules can restrict CI insertion while still allowing a source to update CIs that another trusted process created. This is useful when a feed has valuable attributes but weak coverage or identity.

The combination is important. A source may be trusted to enrich a server CI with vulnerability context but not trusted to create server CIs on its own. Separating create authority from attribute update authority gives the CMDB a more precise trust model than an all-or-nothing integration permission.

Insertion restrictions can protect classes that must be created only through a trusted lifecycle. For example, an enrichment source can update vulnerability or owner fields on server CIs without being allowed to create servers from partial telemetry. This keeps the source useful while preserving the class population under stronger discovery control.

Reconciliation failures often appear as value oscillation

When two sources repeatedly overwrite one another, operators see fields that change back and forth after each scheduled job. Another symptom is a field that never updates even though the source clearly contains a newer value. Both patterns point to authority decisions, not necessarily connector failure.

CMDB health should include source-level diagnostics so teams can distinguish stale data from reconciliation skips. A skipped write may be correct policy. The monitoring question is whether the field value matches the source that the organization intentionally declared authoritative.

A value that never changes can be just as dangerous as an oscillating one. If the authoritative source was removed or renamed, a reconciliation rule may continue blocking all remaining writers, leaving the field stale indefinitely. Monitoring should therefore flag important attributes whose last successful update is much older than the rest of the CI.

Reconciliation logs and IRE diagnostics should be retained long enough to investigate slow-moving quality problems. If an attribute was rejected several weeks ago and operators only notice when a report is wrong, short diagnostic retention can make the reason impossible to reconstruct. Operational retention should match the cadence at which important CMDB data is reviewed.

Manual updates need an explicit place in the authority model

Manual entry is often treated as inherently trustworthy because a human made the change. That assumption can be dangerous. An operator may know a business attribute better than Discovery, while a manually typed technical attribute may become stale immediately. Reconciliation should reflect the nature of the field, not the prestige of the source.

Organizations should decide where human ownership is legitimate and how exceptions expire. If a manual override is temporary, a process should return authority to the automated source later. Permanent overrides should be visible, owned, and reviewed so they do not become undocumented reasons that automation can no longer correct a CI.

Manual exceptions should have expiration logic where possible. A human may need to override an automated value during a migration or incident, but the exception should not silently become permanent ownership. Recording an owner and review date prevents temporary reconciliation policy from turning into hidden long-term data debt.

Test rule changes with competing payloads before production

A reconciliation policy is difficult to validate with a single-source test. Teams should send controlled payloads from two or more discovery sources to the same CI and verify that authorized attributes change while protected attributes remain stable. Test cases should include nulls, stale values, and partial records.

Regression testing should also cover rule priority and class inheritance. A definition that works for a parent class may have unintended behavior on a child class. Engineers need evidence that a rule change does not silently widen or narrow authority across a large class hierarchy.

Governance should answer who can change the authority model

Reconciliation rules are not just technical configuration; they encode organizational trust. Changing them can transfer control of an attribute from one team or system to another. That decision should have an owner, rationale, impact analysis, and rollback plan.

CMDB governance provides the operating structure for those choices. Data owners should define business authority, platform engineers should implement and test the rules, and source owners should understand what their feed is permitted to change. When those responsibilities are clear, reconciliation stops being a hidden IRE behavior and becomes a deliberate data contract.

Authority changes should be reviewed with downstream consumers in mind. If a field drives routing, service impact, vulnerability ownership, or automation, changing its source can alter behavior beyond the CMDB form. The reconciliation change should therefore include a list of critical consumers and validation that they still receive the expected values.

Leave a Reply

How It Works

img
Step 1. Choose Exam
on ExamLabs
Download IT Exams Questions & Answers
img
Step 2. Open Exam with
Avanset Exam Simulator
Press here to download VCE Exam Simulator that simulates real exam environment
img
Step 3. Study
& Pass
IT Exams Anywhere, Anytime!