A web application uses an Application Load Balancer (ALB) to route traffic to EC2 instances across three Availability Zones. A new application version was deployed to one Availability Zone for testing. If a problem occurs, the company needs to route traffic away from that Availability Zone until the deployment is rolled back, while keeping the application available and stable. Which approach provides the required behavior with the most operational efficiency?

The correct answer is to disable cross-zone load balancing on the ALB target group and initiate a zonal shift. Disabling cross-zone load balancing ensures that the ALB only sends traffic to instances within the same Availability Zone as the client. This is crucial for isolating the problem. Initiating a zonal shift then allows you to quickly and safely remove traffic from the problematic Availability Zone without impacting the other zones, maintaining application availability. Manually removing instances is less efficient and prone to human error. Configuring the target group to inherit the ALB's cross-zone load balancing setting would continue to send traffic to the problematic zone, defeating the purpose of isolation. Removing the subnet associated with the affected Availability Zone is a drastic measure that would likely cause more disruption than a zonal shift.