An ecommerce team wants to deploy a new SageMaker model behind their existing real-time recommendation API without changing client applications, and they want to test the new model on a portion of production traffic before full rollout. Which approach provides this with the least operational overhead?

SageMaker production variants allow you to deploy multiple models behind a single endpoint and distribute incoming inference requests among them based on a specified traffic weight. This enables A/B testing, blue/green deployments, and gradual rollouts without changing the client application's API endpoint, minimizing operational overhead. Creating new endpoints with an ALB or NLB would require managing external load balancers and potentially updating client configurations or an API Gateway, increasing complexity. SageMaker batch transform is for offline, asynchronous inference on large datasets, not real-time API traffic routing.