Question 300
A decommissioned application no longer uses a production model, and storage costs grow with obsolete artifacts. What cleanup strategy is appropriate?
Explanation
Correct Answer
Safe lifecycle cleanup balances operational cost with rollback needs and legally required artifact retention.
Incorrect Answers
Maintain accountable owner and lifecycle status metadata for every registered model family. This is useful for the distinct issue: A data science team has no record of who owns models abandoned in a shared registry. The relevant technical rationale or evidence for that issue is: Ownership information enables safe review, maintenance and retirement decisions when model-producing staff or projects change. Here the reported issue is: A decommissioned application no longer uses a production model, and storage costs grow with obsolete artifacts. The relevant response is: Archive or remove obsolete model versions only after dependency and retention checks. The first result cannot verify this different correction: Safe lifecycle cleanup balances operational cost with rollback needs and legally required artifact retention.
Record the deprecation reason, reviewer decision, and replacement version alongside lifecycle state. It addresses another case: A team needs to show why an older model version stopped serving due to fairness concerns. For that case, the distinguishing justification or acceptance check is: Governance metadata explains retirement decisions and prevents unsafe unreviewed reactivation of an old artifact. The present case instead states: A decommissioned application no longer uses a production model, and storage costs grow with obsolete artifacts. It calls for Archive or remove obsolete model versions only after dependency and retention checks. Showing success for the other case would not establish: Safe lifecycle cleanup balances operational cost with rollback needs and legally required artifact retention.
The alternative, Use a stable registered model name and explicit version rather than transient storage paths, belongs to a different technical condition: A production rollout references a model by its environment-specific storage URI, which changes after a workspace migration. There the source-specific mechanism or proof is: Registry identifiers separate logical model identity from storage location, improving portability across controlled deployments. The condition asked about here is: A decommissioned application no longer uses a production model, and storage costs grow with obsolete artifacts. A suitable correction is Archive or remove obsolete model versions only after dependency and retention checks; the targeted justification or result is: Safe lifecycle cleanup balances operational cost with rollback needs and legally required artifact retention.