Certified Machine Learning ProfessionalDemo

By Databricks · Question Mode

QuestionQ72

Deployment Strategies

A Machine Learning Engineer has a production AI model that predicts fraudulent transactions in real time. The model is integrated into critical business workflows, and any outage could lead to significant financial loss and regulatory penalties. The engineer must deploy a new, improved model version to production to replace the current model. The bank’s fairness policy requires all transactions to be exposed to the same model at any given time. Additionally, the deployment must satisfy these requirements:

  • Zero downtime: The fraud-detection service must stay continuously available to users and downstream systems.
  • Immediate rollback: If the new model causes problems, the previous version must be restored instantly.

Which deployment strategy satisfies these requirements?

  • A Canary
  • B Rolling
  • C Shadow
  • D Blue-Green
Explanation

A blue-green deployment runs the existing and new model versions in separate production environments, then switches all production traffic to the new environment at once. This preserves continuous availability and ensures transactions are served by a single model version at a given time. Keeping the previous environment running allows traffic to be switched back rapidly if the new version causes issues.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!