The root cause of the issue is a lack of high availability in the inspection fleet. With only one inspection instance per Availability Zone (AZ), maintenance on that instance creates a single point of failure for all traffic originating within that AZ. By default, a Gateway Load Balancer (GWLB) only sends traffic to targets within the same AZ.

Deploying an additional inspection instance in each AZ (B) provides intra-AZ redundancy, ensuring that if one instance is down for maintenance, the other can process the traffic.

Enabling cross-zone load balancing (C) provides inter-AZ redundancy. If all inspection instances in one AZ become unhealthy, the GWLB will route traffic to healthy instances in other enabled AZs, preventing session timeouts.

A. Deploying instances in other, unused AZs does not solve the single point of failure within the two AZs where the application and traffic originate.

D. A CPU-based scaling policy addresses performance, not the availability failure described during maintenance. The problem is an insufficient number of instances, not high load.

E. Attaching the GWLB to more AZs is ineffective without also deploying target instances in them or enabling cross-zone load balancing to use targets in other AZs.

1. Elastic Load Balancing User Guide - Cross-zone load balancing for Gateway Load Balancers: "By default, each Gateway Load Balancer node distributes traffic only to the registered targets in its Availability Zone. If you enable cross-zone load balancing, each Gateway Load Balancer node distributes traffic to the registered targets in all enabled Availability Zones." This directly supports option C as a solution for AZ-level target failure.

Source: AWS Documentation, Elastic Load Balancing User Guide, "Gateway Load Balancers", Section: "Cross-zone load balancing".

2. AWS Well-Architected Framework - Reliability Pillar: "Withstand component failure: Ensure that you have a strategy to withstand component failure. For example, you can use redundant components..." Deploying a second inspection instance in each AZ (Option B) is a direct implementation of this principle to achieve high availability.

Source: AWS Well-Architected Framework, Reliability Pillar, Design Principle: "REL 3: How do you design your workload architecture to be highly available?", Section: "Withstand component failure".

3. AWS Blog - Scaling network traffic inspection using AWS Gateway Load Balancer: "For high availability, we recommend deploying the security appliances in an Auto Scaling group, distributed across multiple Availability Zones. You can also enable cross-zone load balancing on your Gateway Load Balancer to distribute traffic across appliances in all Availability Zones." This reference explicitly recommends both deploying multiple appliances per AZ (achieved by Option B) and enabling cross-zone load balancing (Option C) for high availability.

Source: AWS Networking & Content Delivery Blog, "Scaling network traffic inspection using AWS Gateway Load Balancer", November 10, 2020.