Your company's user-feedback portal comprises a standard LAMP stack replicated across two zones. It is deployed in the us-central1 region and uses autoscaled managed instance groups on all layers, except the database. Currently, only a small group of select customers have access to the portal. The portal meets a 99,99% availability SLA under these conditions. However next quarter, your company will be making the portal available to all users, including unauthenticated users. You need to develop a resiliency testing strategy to ensure the system maintains the SLA once they introduce additional user load. What should you do?

The correct option involves creating synthetic random user input and replaying this load to trigger autoscaling. This approach allows for controlled testing without impacting real users. Introducing "chaos" by terminating random resources in both zones is a key aspect of resiliency testing, simulating failures to ensure the system can recover and maintain the SLA. The other options are less effective or carry higher risks. Capturing and replaying existing user input might not accurately represent the new, larger, and potentially unauthenticated user base. Terminating all resources in one zone is too aggressive for initial testing and doesn't simulate random failures. Exposing the new system to a larger group of users incrementally is a form of canary release, not a comprehensive resiliency testing strategy, and terminating random resources on a live system carries significant risk. Deploying resources for 200% of expected load without testing resiliency doesn't guarantee the SLA under failure conditions.