Skip to main content
Gathering Interest

Ability to Identify Deletions from PI-AF Source in CDF

Related products:Extractors
  • March 10, 2026
  • 0 replies
  • 5 views

Forum|alt.badge.img+7

Currently, the PI-AF Extractor works as a forward synchronization tool that reads the PI-AF structure and sends it to Cognite Data Fusion (CDF).

The extractor correctly handles:

  • Insertions: New elements created in PI-AF are detected and created in CDF.

  • Modifications: Updates to elements or attributes (such as renaming elements or moving branches) are synchronized and reflected in CDF.

However, when elements or branches are deleted in PI-AF, the corresponding objects remain in CDF, and there is no indicator that the element no longer exists in the source system. This behavior is currently intentional to prevent accidental or irreversible data loss in CDF.

While this approach protects data in CDF, it creates challenges for downstream processes. Without any indication that an object was deleted in the source system, it becomes difficult to automate workflows that depend on identifying obsolete or removed elements.

Requested capability

A mechanism to identify when an element has been deleted from PI-AF, without necessarily deleting the corresponding object in CDF.

For example, the extractor could mark such objects with a metadata field, tag, or status indicating that the element was removed from the source system.

This would allow users to implement their own logic for handling these cases (e.g., archiving, cleaning up, or removing the objects in downstream processes) while maintaining the current protection against unintended data deletion.

Use case

Organizations that synchronize large PI-AF hierarchies to CDF often rely on automated pipelines and asset-based workflows. When assets or branches are removed from PI-AF, there is currently no straightforward way to detect this change in CDF, making it difficult to automate lifecycle management of these elements.

Providing an indicator that an element was deleted in the source system would enable users to build controlled automation around these events.
 

Regards Daniel