An application runs on EC2 in a single VPC across two Availability Zones. A shared inspection VPC with a GWLB and a fleet of marketplace inspection instances inspects traffic between the VPC and the internet. To save cost, only one inspection instance was deployed in each AZ used by the application. During maintenance of an inspection instance, some application instances experience internet session timeouts and cannot establish new sessions. Which combination of changes will fix the problem? (Choose two.)

The problem describes session timeouts and failures when an inspection instance undergoes maintenance, indicating a lack of high availability and resilience. Deploying one additional inspection instance in each Availability Zone (AZ) where instances are already deployed increases the number of active instances per AZ. This ensures that if one instance becomes unavailable (e.g., for maintenance), another is still available to process traffic, preventing service disruption. Enabling the cross-zone load balancing attribute for the GWLB allows traffic to be distributed across inspection instances in all enabled AZs, even if the GWLB endpoint is in a different AZ. This improves resilience by ensuring that if an entire AZ's inspection instances become unavailable, traffic can still be routed to instances in other AZs, maintaining session continuity. Deploying instances in AZs without them or attaching the GWLB to all AZs in the Region are not direct solutions to the in-AZ instance failure. An Auto Scaling group helps with dynamic scaling but doesn't inherently solve the single point of failure during maintenance without sufficient initial capacity.