MDM Reference Data
Reference data is the controlled vocabulary layer of DFS MDM. Use it for small, governed code lists that should be consistent across source mappings, fusion tasks, validation rules, and reporting.
Examples include severity, status, unit, fault category, equipment class, region, and other controlled values. Real-world objects with source-system aliases belong in master entities.
Before you start
Confirm:
dfs:readaccess for viewing reference sets;dfs:writeaccess for creating or editing sets and values;dfs:deleteaccess only when deleting values is allowed by project policy;- the steward or owner for the code list;
- whether existing downstream datasets or reports depend on the current values.
Open Reference Data
Go to:
Data Integration > DFS Pro > Reference Data
The page shows reference sets on the left and the selected set's values on the right.
Lifecycle
Treat reference data as a shared contract. A value that looks like a small list change can affect validation rules, source mappings, reports, AI Agent evidence, and steward review screens.
Create a reference set
- Select New set.
- Enter a stable key.
- Enter a display name that users can recognize.
- Save the set.
- Select the set and add values.
Use stable keys that describe the business domain and remain valid across environments.
Good examples:
equipment_statusfault_severityinspection_resultenergy_unit
Example value set
For a fault severity list, define values with clear meanings:
| Code | Label | Intended use |
|---|---|---|
critical | Critical | Immediate operational impact or safety concern. |
major | Major | Significant degradation that needs planned response. |
minor | Minor | Low-impact condition for monitoring or routine work. |
info | Information | Context-only event or informational signal. |
Avoid overlapping values such as urgent, critical, and high unless the operations team can explain the difference and apply it consistently.
Add values
For each value, define:
| Field | Use |
|---|---|
| Code | Stable value stored or exchanged by integrations. |
| Label | Human-readable label shown to users. |
| Valid from | First date when the value is accepted. |
| Valid to | Last date when the value is accepted. Leave open for current values. |
| Order | Display order in lists or review screens. |
| Attributes | Optional structured details used by implementation-specific workflows. |
Version a reference set
Use Bump version when a change affects interpretation, downstream validation, or reporting.
Bump the version when:
- a value is added for a new accepted business state;
- a value is retired or replaced;
- labels change in a way that affects reviewers;
- validity dates change;
- downstream datasets or methods need a visible change boundary.
For minor typo fixes, record the reason in the handoff notes and keep the version policy consistent with the project.
Change impact review
Before changing a live reference set, check:
| Impact area | Review question |
|---|---|
| Source mappings | Which imported values map to this code today? |
| DFS Pro methods | Do fusion tasks or validation rules filter by this code? |
| Steward screens | Will reviewers see a different label or option? |
| Reports | Will historical counts move between categories? |
| AI Agent evidence | Does the Agent retrieve, group, or explain records by this code? |
| External integrations | Do partner systems expect the old code or label? |
If a code is retired, keep a historical validity window instead of deleting it when old records still need to validate.
Review checklist
Before publishing changes:
- Codes are stable and have distinct meanings.
- Labels are clear for operators, stewards, and report users.
- Validity periods keep active values unambiguous.
- Active mappings have been checked for deleted or retired values.
- The owner knows which datasets, methods, reports, or AI workflows depend on the set.
Common issues
| Symptom | Likely cause | Fix |
|---|---|---|
| A value is missing in a downstream form | The reference set is stale or the user is looking at another tenant or environment. | Confirm tenant, set key, and version. |
| A report groups values incorrectly | Two codes are being used for the same meaning. | Consolidate the code list and review downstream mappings. |
| Historical records appear invalid | Validity dates were changed without checking historical data. | Restore the historical window or add a replacement value with the right period. |
| Users create free-text variants | A workflow still allows unmanaged values. | Update mapping, form, or method configuration to use the controlled values. |
Next step
Use Master Entities when the data item is a real-world object that needs aliases, lineage, and stewardship.