Skip to main content

Understanding Output

DataGen produces three important categories of output:

Generated world object​

The in-memory generation result is the richest form of output. It contains:

  • the generated world graph
  • generation statistics
  • warnings and diagnostics
  • plugin output, when plugins are enabled

This is the best form to use inside PowerShell pipelines.

Snapshot​

Save-SEEnterpriseWorld writes a portable snapshot that can later be loaded with Import-SEEnterpriseWorld. In v0.9.3, the snapshot provenance is portable: it describes the generated world and supplied source context without treating the local checkout path as its identity.

Use snapshots when you want:

  • repeatable inspection later
  • transport between environments
  • a stable checkpoint before downstream processing

Normalized export​

Export-SEEnterpriseWorld writes flattened artifacts intended for analysis, integration, seeding, and downstream mapping.

Normalized export is the best choice when:

  • another tool needs files rather than a .NET object graph
  • you want deterministic tables or JSON artifacts
  • you are building a bridge or adapter for a consumer system

For a repeatable export, provide a fixed -ExportedAtUtc value and include the manifest. v0.9.3 writes the artifact set in canonical ordinal order with normalized newlines across supported platforms. The resulting manifest.json is the single export identity to retain with the artifact folder; it is not a replacement for keeping the scenario, catalog, and source version that produced it.

Reproducibility prerequisites​

The same seed alone is not enough to make an equality claim. For two independently generated outputs, retain and use the same:

  • scenario content
  • seed
  • generation, snapshot, and export timestamps
  • DataGen version and source identity
  • catalog content
  • public generation arguments and export profile

When those inputs are fixed, the deterministic-generation receipt tooling can compare complete payload inventories. It deliberately excludes declared sensitive inputs from its equality proof. Outputs that intentionally generate credential material can therefore differ even when the non-sensitive world and export contract are the same.

What realism means in practice​

Generated output is not meant to be perfectly neat unless you ask for it. Depending on the scenario and deviation profile, you may see:

  • stale identities
  • inconsistent CMDB ownership
  • partial classifications
  • missing or conflicting policy data
  • realistic access evidence and exceptions

This is intentional. The goal is a believable environment, not an already-clean one.

Canonical vs observed vs exported​

For newer surfaces such as CMDB and observed data, it helps to think in layers:

  • canonical world truth
  • observed or source-system views
  • normalized export artifacts built from those layers

That separation lets DataGen produce both the “actual world” and the imperfect records a real organization might maintain about it.

Next step​

If you want more control over what gets generated, continue to Scenario Authoring.