Question

Your team runs a managed online endpoint serving a fraud-detection model with a stable deployment named 'blue' handling 100% of production traffic. You have trained an improved model and created a new deployment named 'green' on the same endpoint. Company policy requires that new models be validated against a small slice of live production traffic before receiving full rollout, and that you retain the ability to instantly revert if error rates spike. Which action best satisfies this requirement while minimizing production risk?

Accepted Answer

Update the endpoint traffic allocation so that 'blue' receives 90% and 'green' receives 10%, then monitor 'green' metrics before increasing its share.

Answer

Delete the 'blue' deployment and route 100% of traffic to 'green' immediately to complete the upgrade in a single step.

Answer

Deploy 'green' to a separate endpoint and manually redirect a copy of production requests to it using a client-side script.

Answer

Set the 'green' deployment's instance count higher than 'blue' so it automatically absorbs the majority of incoming traffic.