A ticketing application runs on an Amazon ECS cluster and is delivered via CloudFront. A single ECS service is the CloudFront origin. The app limits the number of active users in the purchase flow; allowed users are identified by an encrypted attribute in their JWT; others are routed to a waiting room. Under high load, the waiting room itself is causing load that disrupts ticket purchases. What change will provide the most reliable ticket purchase experience during high load?

The correct answer isolates the waiting room's load from the main ticketing application. By creating a separate ECS service for the waiting room, it can scale independently, preventing its load from impacting ticket purchases. A CloudFront Function is ideal for inspecting the JWT at the edge and routing requests to the appropriate service (ticketing or waiting room) before they even reach the origin, minimizing unnecessary load on the ticketing service. Option 1 is incorrect because having the ticketing service forward requests means the ticketing service still processes all requests, even those destined for the waiting room, which doesn't fully alleviate its load. Option 2 and 4 involve migrating to EKS, which is a significant architectural change and not the most direct or efficient solution to the immediate problem of separating load. Furthermore, App Mesh (Option 4) adds complexity without directly solving the initial routing problem at the edge.